diff --git a/camelot/handlers.py b/camelot/handlers.py index f640c1b..59b31c3 100644 --- a/camelot/handlers.py +++ b/camelot/handlers.py @@ -81,6 +81,8 @@ class PDFHandler(object): """ with open(filename, 'rb') as fileobj: infile = PdfFileReader(fileobj, strict=False) + if infile.isEncrypted: + infile.decrypt('') fpath = os.path.join(temp, 'page-{0}.pdf'.format(page)) froot, fext = os.path.splitext(fpath) p = infile.getPage(page - 1) @@ -98,6 +100,8 @@ class PDFHandler(object): fpath_new = ''.join([froot.replace('page', 'p'), '_rotated', fext]) os.rename(fpath, fpath_new) infile = PdfFileReader(open(fpath_new, 'rb'), strict=False) + if infile.isEncrypted: + infile.decrypt('') outfile = PdfFileWriter() p = infile.getPage(0) if rotation == 'anticlockwise': diff --git a/camelot/parsers/stream.py b/camelot/parsers/stream.py index 56a240d..6d29a05 100644 --- a/camelot/parsers/stream.py +++ b/camelot/parsers/stream.py @@ -294,8 +294,7 @@ class Stream(BaseParser): if ncols == 1: logger.info("No tables found on {}".format( os.path.basename(self.rootname))) - cols = [(t.x0, t.x1) - for r in rows_grouped if len(r) == ncols for t in r] + cols = [(t.x0, t.x1) for r in rows_grouped if len(r) == ncols for t in r] cols = self._merge_columns(sorted(cols), col_close_tol=self.col_close_tol) inner_text = [] for i in range(1, len(cols)): diff --git a/docs/benchmark/agstat/agstat-data-camelot-page-1-table-1.csv b/docs/benchmark/lattice/agstat/agstat-data-camelot-page-1-table-1.csv similarity index 100% rename from docs/benchmark/agstat/agstat-data-camelot-page-1-table-1.csv rename to docs/benchmark/lattice/agstat/agstat-data-camelot-page-1-table-1.csv diff --git a/docs/benchmark/agstat/agstat-data-tabula.csv b/docs/benchmark/lattice/agstat/agstat-data-tabula.csv similarity index 100% rename from docs/benchmark/agstat/agstat-data-tabula.csv rename to docs/benchmark/lattice/agstat/agstat-data-tabula.csv diff --git a/docs/benchmark/agstat/agstat-table-detection-camelot.png b/docs/benchmark/lattice/agstat/agstat-table-detection-camelot.png similarity index 100% rename from docs/benchmark/agstat/agstat-table-detection-camelot.png rename to docs/benchmark/lattice/agstat/agstat-table-detection-camelot.png diff --git a/docs/benchmark/agstat/agstat-table-detection-tabula.png b/docs/benchmark/lattice/agstat/agstat-table-detection-tabula.png similarity index 100% rename from docs/benchmark/agstat/agstat-table-detection-tabula.png rename to docs/benchmark/lattice/agstat/agstat-table-detection-tabula.png diff --git a/docs/benchmark/agstat/agstat.pdf b/docs/benchmark/lattice/agstat/agstat.pdf similarity index 100% rename from docs/benchmark/agstat/agstat.pdf rename to docs/benchmark/lattice/agstat/agstat.pdf diff --git a/docs/benchmark/background_lines_1/background_lines_1-data-camelot-page-1-table-1.csv b/docs/benchmark/lattice/background_lines_1/background_lines_1-data-camelot-page-1-table-1.csv similarity index 100% rename from docs/benchmark/background_lines_1/background_lines_1-data-camelot-page-1-table-1.csv rename to docs/benchmark/lattice/background_lines_1/background_lines_1-data-camelot-page-1-table-1.csv diff --git a/docs/benchmark/background_lines_1/background_lines_1-data-camelot-page-1-table-2.csv b/docs/benchmark/lattice/background_lines_1/background_lines_1-data-camelot-page-1-table-2.csv similarity index 100% rename from docs/benchmark/background_lines_1/background_lines_1-data-camelot-page-1-table-2.csv rename to docs/benchmark/lattice/background_lines_1/background_lines_1-data-camelot-page-1-table-2.csv diff --git a/docs/benchmark/background_lines_1/background_lines_1-data-tabula.csv b/docs/benchmark/lattice/background_lines_1/background_lines_1-data-tabula.csv similarity index 100% rename from docs/benchmark/background_lines_1/background_lines_1-data-tabula.csv rename to docs/benchmark/lattice/background_lines_1/background_lines_1-data-tabula.csv diff --git a/docs/benchmark/background_lines_1/background_lines_1-table-detection-camelot.png b/docs/benchmark/lattice/background_lines_1/background_lines_1-table-detection-camelot.png similarity index 100% rename from docs/benchmark/background_lines_1/background_lines_1-table-detection-camelot.png rename to docs/benchmark/lattice/background_lines_1/background_lines_1-table-detection-camelot.png diff --git a/docs/benchmark/background_lines_1/background_lines_1-table-detection-tabula.png b/docs/benchmark/lattice/background_lines_1/background_lines_1-table-detection-tabula.png similarity index 100% rename from docs/benchmark/background_lines_1/background_lines_1-table-detection-tabula.png rename to docs/benchmark/lattice/background_lines_1/background_lines_1-table-detection-tabula.png diff --git a/docs/benchmark/background_lines_1/background_lines_1.pdf b/docs/benchmark/lattice/background_lines_1/background_lines_1.pdf similarity index 100% rename from docs/benchmark/background_lines_1/background_lines_1.pdf rename to docs/benchmark/lattice/background_lines_1/background_lines_1.pdf diff --git a/docs/benchmark/background_lines_2/background_lines_2-data-camelot-page-1-table-1.csv b/docs/benchmark/lattice/background_lines_2/background_lines_2-data-camelot-page-1-table-1.csv similarity index 100% rename from docs/benchmark/background_lines_2/background_lines_2-data-camelot-page-1-table-1.csv rename to docs/benchmark/lattice/background_lines_2/background_lines_2-data-camelot-page-1-table-1.csv diff --git a/docs/benchmark/background_lines_2/background_lines_2-data-tabula.csv b/docs/benchmark/lattice/background_lines_2/background_lines_2-data-tabula.csv similarity index 100% rename from docs/benchmark/background_lines_2/background_lines_2-data-tabula.csv rename to docs/benchmark/lattice/background_lines_2/background_lines_2-data-tabula.csv diff --git a/docs/benchmark/background_lines_2/background_lines_2-table-detection-camelot.png b/docs/benchmark/lattice/background_lines_2/background_lines_2-table-detection-camelot.png similarity index 100% rename from docs/benchmark/background_lines_2/background_lines_2-table-detection-camelot.png rename to docs/benchmark/lattice/background_lines_2/background_lines_2-table-detection-camelot.png diff --git a/docs/benchmark/background_lines_2/background_lines_2-table-detection-tabula.png b/docs/benchmark/lattice/background_lines_2/background_lines_2-table-detection-tabula.png similarity index 100% rename from docs/benchmark/background_lines_2/background_lines_2-table-detection-tabula.png rename to docs/benchmark/lattice/background_lines_2/background_lines_2-table-detection-tabula.png diff --git a/docs/benchmark/background_lines_2/background_lines_2.pdf b/docs/benchmark/lattice/background_lines_2/background_lines_2.pdf similarity index 100% rename from docs/benchmark/background_lines_2/background_lines_2.pdf rename to docs/benchmark/lattice/background_lines_2/background_lines_2.pdf diff --git a/docs/benchmark/column_span_1/column_span_1-data-camelot-page-1-table-1.csv b/docs/benchmark/lattice/column_span_1/column_span_1-data-camelot-page-1-table-1.csv similarity index 100% rename from docs/benchmark/column_span_1/column_span_1-data-camelot-page-1-table-1.csv rename to docs/benchmark/lattice/column_span_1/column_span_1-data-camelot-page-1-table-1.csv diff --git a/docs/benchmark/column_span_1/column_span_1-data-tabula.csv b/docs/benchmark/lattice/column_span_1/column_span_1-data-tabula.csv similarity index 100% rename from docs/benchmark/column_span_1/column_span_1-data-tabula.csv rename to docs/benchmark/lattice/column_span_1/column_span_1-data-tabula.csv diff --git a/docs/benchmark/column_span_1/column_span_1-table-detection-camelot.png b/docs/benchmark/lattice/column_span_1/column_span_1-table-detection-camelot.png similarity index 100% rename from docs/benchmark/column_span_1/column_span_1-table-detection-camelot.png rename to docs/benchmark/lattice/column_span_1/column_span_1-table-detection-camelot.png diff --git a/docs/benchmark/column_span_1/column_span_1-table-detection-tabula.png b/docs/benchmark/lattice/column_span_1/column_span_1-table-detection-tabula.png similarity index 100% rename from docs/benchmark/column_span_1/column_span_1-table-detection-tabula.png rename to docs/benchmark/lattice/column_span_1/column_span_1-table-detection-tabula.png diff --git a/docs/benchmark/column_span_1/column_span_1.pdf b/docs/benchmark/lattice/column_span_1/column_span_1.pdf similarity index 100% rename from docs/benchmark/column_span_1/column_span_1.pdf rename to docs/benchmark/lattice/column_span_1/column_span_1.pdf diff --git a/docs/benchmark/column_span_2/column_span_2-data-camelot-page-1-table-1.csv b/docs/benchmark/lattice/column_span_2/column_span_2-data-camelot-page-1-table-1.csv similarity index 100% rename from docs/benchmark/column_span_2/column_span_2-data-camelot-page-1-table-1.csv rename to docs/benchmark/lattice/column_span_2/column_span_2-data-camelot-page-1-table-1.csv diff --git a/docs/benchmark/column_span_2/column_span_2-data-tabula.csv b/docs/benchmark/lattice/column_span_2/column_span_2-data-tabula.csv similarity index 100% rename from docs/benchmark/column_span_2/column_span_2-data-tabula.csv rename to docs/benchmark/lattice/column_span_2/column_span_2-data-tabula.csv diff --git a/docs/benchmark/column_span_2/column_span_2-table-detection-camelot.png b/docs/benchmark/lattice/column_span_2/column_span_2-table-detection-camelot.png similarity index 100% rename from docs/benchmark/column_span_2/column_span_2-table-detection-camelot.png rename to docs/benchmark/lattice/column_span_2/column_span_2-table-detection-camelot.png diff --git a/docs/benchmark/column_span_2/column_span_2-table-detection-tabula.png b/docs/benchmark/lattice/column_span_2/column_span_2-table-detection-tabula.png similarity index 100% rename from docs/benchmark/column_span_2/column_span_2-table-detection-tabula.png rename to docs/benchmark/lattice/column_span_2/column_span_2-table-detection-tabula.png diff --git a/docs/benchmark/column_span_2/column_span_2.pdf b/docs/benchmark/lattice/column_span_2/column_span_2.pdf similarity index 100% rename from docs/benchmark/column_span_2/column_span_2.pdf rename to docs/benchmark/lattice/column_span_2/column_span_2.pdf diff --git a/docs/benchmark/electoral_roll/electoral_roll-data-camelot-page-1-table-1.csv b/docs/benchmark/lattice/electoral_roll/electoral_roll-data-camelot-page-1-table-1.csv similarity index 100% rename from docs/benchmark/electoral_roll/electoral_roll-data-camelot-page-1-table-1.csv rename to docs/benchmark/lattice/electoral_roll/electoral_roll-data-camelot-page-1-table-1.csv diff --git a/docs/benchmark/electoral_roll/electoral_roll-data-tabula.csv b/docs/benchmark/lattice/electoral_roll/electoral_roll-data-tabula.csv similarity index 100% rename from docs/benchmark/electoral_roll/electoral_roll-data-tabula.csv rename to docs/benchmark/lattice/electoral_roll/electoral_roll-data-tabula.csv diff --git a/docs/benchmark/electoral_roll/electoral_roll-table-detection-camelot.png b/docs/benchmark/lattice/electoral_roll/electoral_roll-table-detection-camelot.png similarity index 100% rename from docs/benchmark/electoral_roll/electoral_roll-table-detection-camelot.png rename to docs/benchmark/lattice/electoral_roll/electoral_roll-table-detection-camelot.png diff --git a/docs/benchmark/electoral_roll/electoral_roll-table-detection-tabula.png b/docs/benchmark/lattice/electoral_roll/electoral_roll-table-detection-tabula.png similarity index 100% rename from docs/benchmark/electoral_roll/electoral_roll-table-detection-tabula.png rename to docs/benchmark/lattice/electoral_roll/electoral_roll-table-detection-tabula.png diff --git a/docs/benchmark/electoral_roll/electoral_roll.pdf b/docs/benchmark/lattice/electoral_roll/electoral_roll.pdf similarity index 100% rename from docs/benchmark/electoral_roll/electoral_roll.pdf rename to docs/benchmark/lattice/electoral_roll/electoral_roll.pdf diff --git a/docs/benchmark/rotated/rotated-data-camelot-page-1-table-1.csv b/docs/benchmark/lattice/rotated/rotated-data-camelot-page-1-table-1.csv similarity index 100% rename from docs/benchmark/rotated/rotated-data-camelot-page-1-table-1.csv rename to docs/benchmark/lattice/rotated/rotated-data-camelot-page-1-table-1.csv diff --git a/docs/benchmark/rotated/rotated-data-tabula.csv b/docs/benchmark/lattice/rotated/rotated-data-tabula.csv similarity index 100% rename from docs/benchmark/rotated/rotated-data-tabula.csv rename to docs/benchmark/lattice/rotated/rotated-data-tabula.csv diff --git a/docs/benchmark/rotated/rotated-table-detection-camelot.png b/docs/benchmark/lattice/rotated/rotated-table-detection-camelot.png similarity index 100% rename from docs/benchmark/rotated/rotated-table-detection-camelot.png rename to docs/benchmark/lattice/rotated/rotated-table-detection-camelot.png diff --git a/docs/benchmark/rotated/rotated-table-detection-tabula.png b/docs/benchmark/lattice/rotated/rotated-table-detection-tabula.png similarity index 100% rename from docs/benchmark/rotated/rotated-table-detection-tabula.png rename to docs/benchmark/lattice/rotated/rotated-table-detection-tabula.png diff --git a/docs/benchmark/rotated/rotated.pdf b/docs/benchmark/lattice/rotated/rotated.pdf similarity index 100% rename from docs/benchmark/rotated/rotated.pdf rename to docs/benchmark/lattice/rotated/rotated.pdf diff --git a/docs/benchmark/row_span/row_span-data-camelot-page-1-table-1.csv b/docs/benchmark/lattice/row_span/row_span-data-camelot-page-1-table-1.csv similarity index 100% rename from docs/benchmark/row_span/row_span-data-camelot-page-1-table-1.csv rename to docs/benchmark/lattice/row_span/row_span-data-camelot-page-1-table-1.csv diff --git a/docs/benchmark/row_span/row_span-data-tabula.csv b/docs/benchmark/lattice/row_span/row_span-data-tabula.csv similarity index 100% rename from docs/benchmark/row_span/row_span-data-tabula.csv rename to docs/benchmark/lattice/row_span/row_span-data-tabula.csv diff --git a/docs/benchmark/row_span/row_span-table-detection-camelot.png b/docs/benchmark/lattice/row_span/row_span-table-detection-camelot.png similarity index 100% rename from docs/benchmark/row_span/row_span-table-detection-camelot.png rename to docs/benchmark/lattice/row_span/row_span-table-detection-camelot.png diff --git a/docs/benchmark/row_span/row_span-table-detection-tabula.png b/docs/benchmark/lattice/row_span/row_span-table-detection-tabula.png similarity index 100% rename from docs/benchmark/row_span/row_span-table-detection-tabula.png rename to docs/benchmark/lattice/row_span/row_span-table-detection-tabula.png diff --git a/docs/benchmark/row_span/row_span.pdf b/docs/benchmark/lattice/row_span/row_span.pdf similarity index 100% rename from docs/benchmark/row_span/row_span.pdf rename to docs/benchmark/lattice/row_span/row_span.pdf diff --git a/docs/benchmark/twotables_1/twotables_1-data-camelot-page-1-table-1.csv b/docs/benchmark/lattice/twotables_1/twotables_1-data-camelot-page-1-table-1.csv similarity index 100% rename from docs/benchmark/twotables_1/twotables_1-data-camelot-page-1-table-1.csv rename to docs/benchmark/lattice/twotables_1/twotables_1-data-camelot-page-1-table-1.csv diff --git a/docs/benchmark/twotables_1/twotables_1-data-camelot-page-1-table-2.csv b/docs/benchmark/lattice/twotables_1/twotables_1-data-camelot-page-1-table-2.csv similarity index 100% rename from docs/benchmark/twotables_1/twotables_1-data-camelot-page-1-table-2.csv rename to docs/benchmark/lattice/twotables_1/twotables_1-data-camelot-page-1-table-2.csv diff --git a/docs/benchmark/twotables_1/twotables_1-data-tabula.csv b/docs/benchmark/lattice/twotables_1/twotables_1-data-tabula.csv similarity index 100% rename from docs/benchmark/twotables_1/twotables_1-data-tabula.csv rename to docs/benchmark/lattice/twotables_1/twotables_1-data-tabula.csv diff --git a/docs/benchmark/twotables_1/twotables_1-table-detection-camelot.png b/docs/benchmark/lattice/twotables_1/twotables_1-table-detection-camelot.png similarity index 100% rename from docs/benchmark/twotables_1/twotables_1-table-detection-camelot.png rename to docs/benchmark/lattice/twotables_1/twotables_1-table-detection-camelot.png diff --git a/docs/benchmark/twotables_1/twotables_1-table-detection-tabula.png b/docs/benchmark/lattice/twotables_1/twotables_1-table-detection-tabula.png similarity index 100% rename from docs/benchmark/twotables_1/twotables_1-table-detection-tabula.png rename to docs/benchmark/lattice/twotables_1/twotables_1-table-detection-tabula.png diff --git a/docs/benchmark/twotables_1/twotables_1.pdf b/docs/benchmark/lattice/twotables_1/twotables_1.pdf similarity index 100% rename from docs/benchmark/twotables_1/twotables_1.pdf rename to docs/benchmark/lattice/twotables_1/twotables_1.pdf diff --git a/docs/benchmark/twotables_2/twotables_2-data-camelot-page-1-table-1.csv b/docs/benchmark/lattice/twotables_2/twotables_2-data-camelot-page-1-table-1.csv similarity index 100% rename from docs/benchmark/twotables_2/twotables_2-data-camelot-page-1-table-1.csv rename to docs/benchmark/lattice/twotables_2/twotables_2-data-camelot-page-1-table-1.csv diff --git a/docs/benchmark/twotables_2/twotables_2-data-camelot-page-1-table-2.csv b/docs/benchmark/lattice/twotables_2/twotables_2-data-camelot-page-1-table-2.csv similarity index 100% rename from docs/benchmark/twotables_2/twotables_2-data-camelot-page-1-table-2.csv rename to docs/benchmark/lattice/twotables_2/twotables_2-data-camelot-page-1-table-2.csv diff --git a/docs/benchmark/twotables_2/twotables_2-table-detection-camelot.png b/docs/benchmark/lattice/twotables_2/twotables_2-table-detection-camelot.png similarity index 100% rename from docs/benchmark/twotables_2/twotables_2-table-detection-camelot.png rename to docs/benchmark/lattice/twotables_2/twotables_2-table-detection-camelot.png diff --git a/docs/benchmark/twotables_2/twotables_2-table-detection-tabula.png b/docs/benchmark/lattice/twotables_2/twotables_2-table-detection-tabula.png similarity index 100% rename from docs/benchmark/twotables_2/twotables_2-table-detection-tabula.png rename to docs/benchmark/lattice/twotables_2/twotables_2-table-detection-tabula.png diff --git a/docs/benchmark/lattice/twotables_2/twotables_2-tabula-0.csv b/docs/benchmark/lattice/twotables_2/twotables_2-tabula-0.csv new file mode 100755 index 0000000..9b653c4 --- /dev/null +++ b/docs/benchmark/lattice/twotables_2/twotables_2-tabula-0.csv @@ -0,0 +1,13 @@ +"",,,Literacy Status, +State,n,Read & Illiterate CollegeWrite,1-4 std. 5-8 std. 9-12 std., +Kerala,2400,7.2 0.5,25.3 20.1 41.5,5.5 +Tamil Nadu,2400,21.4 2.3,8.8 35.5 25.8,6.2 +Karnataka,2399,37.4 2.8,12.5 18.3 23.1,5.8 +Andhra Pradesh,2400,54.0 1.7,8.4 13.2 18.8,3.9 +Maharashtra,2400,22.0 0.9,17.3 20.3 32.6,7.0 +Gujarat,2390,28.6 0.1,14.4 23.1 26.9,6.8 +Madhya Pradesh,2402,29.1 3.4,8.5 35.1 13.3,10.6 +Orissa,2405,33.2 1.0,10.4 25.7 21.2,8.5 +West Bengal,2293,41.7 4.4,13.2 17.1 21.2,2.4 +Uttar Pradesh,2400,35.3 2.1,4.5 23.3 27.1,7.6 +Pooled,23889,30.9 1.9,12.3 23.2 25.2,6.4 diff --git a/docs/benchmark/lattice/twotables_2/twotables_2-tabula-1.csv b/docs/benchmark/lattice/twotables_2/twotables_2-tabula-1.csv new file mode 100755 index 0000000..5322e31 --- /dev/null +++ b/docs/benchmark/lattice/twotables_2/twotables_2-tabula-1.csv @@ -0,0 +1,13 @@ +"",,,Literacy Status, +State,n,Read & Illiterate CollegeWrite,1-4 std. 5-8 std. 9-12 std., +Kerala,2400,8.8 0.3,20.1 17.0 45.6,8.2 +Tamil Nadu,2400,29.9 1.5,8.5 33.1 22.3,4.8 +Karnataka,2399,47.9 2.5,10.2 18.8 18.4,2.3 +Andhra Pradesh,2400,66.4 0.7,6.8 12.9 11.4,1.8 +Maharashtra,2400,41.3 0.6,14.1 20.1 21.6,2.2 +Gujarat,2390,57.6 0.1,10.3 16.5 12.9,2.7 +Madhya Pradesh,2402,58.7 2.2,6.6 24.1 5.3,3.0 +Orissa,2405,50.0 0.9,8.1 21.9 15.1,4.0 +West Bengal,2293,49.1 4.8,11.2 16.8 17.1,1.1 +Uttar Pradesh,2400,67.3 2.0,3.1 17.2 7.7,2.7 +Pooled,23889,47.7 1.5,9.9 19.9 17.8,3.3 diff --git a/docs/benchmark/twotables_2/twotables_2.pdf b/docs/benchmark/lattice/twotables_2/twotables_2.pdf similarity index 100% rename from docs/benchmark/twotables_2/twotables_2.pdf rename to docs/benchmark/lattice/twotables_2/twotables_2.pdf diff --git a/docs/benchmark/stream/12s0324/12s0324-data-camelot-page-1-table-1.csv b/docs/benchmark/stream/12s0324/12s0324-data-camelot-page-1-table-1.csv new file mode 100755 index 0000000..a8cffd6 --- /dev/null +++ b/docs/benchmark/stream/12s0324/12s0324-data-camelot-page-1-table-1.csv @@ -0,0 +1,38 @@ +"","","Total","","","Male","","","Female","" +"Offense charged","","Under 18","18 years","","Under 18","18 years","","Under 18","18 years" +"","Total","years","and over","Total","years","and over","Total","years","and over" +"Total . . . . . . . . . . . . . . . . . . . . . . . . .","11,062 .6","1,540 .0","9,522 .6","8,263 .3","1,071 .6","7,191 .7","2,799 .2","468 .3","2,330 .9" +"Violent crime . . . . . . . . . . . . . . . . . .","467 .9","69 .1","398 .8","380 .2","56 .5","323 .7","87 .7","12 .6","75 .2" +"Murder and nonnegligent","","","","","","","","","" +"manslaughter . . . . . . . .. .. .. .. ..","10.0","0.9","9.1","9.0","0.9","8.1","1.1","–","1.0" +"Forcible rape . . . . . . . .. .. .. .. .. .","17.5","2.6","14.9","17.2","2.5","14.7","–","–","–" +"Robbery . . . .. .. . .. . ... . ... . ...","102.1","25.5","76.6","90.0","22.9","67.1","12.1","2.5","9.5" +"Aggravated assault . . . . . . . .. .. ..","338.4","40.1","298.3","264.0","30.2","233.8","74.4","9.9","64.5" +"Property crime . . . . . . . . . . . . . . . . .","1,396 .4","338 .7","1,057 .7","875 .9","210 .8","665 .1","608 .2","127 .9","392 .6" +"Burglary . .. . . . . .. ... .... .... ..","240.9","60.3","180.6","205.0","53.4","151.7","35.9","6.9","29.0" +"Larceny-theft . . . . . . . .. .. .. .. .. .","1,080.1","258.1","822.0","608.8","140.5","468.3","471.3","117.6","353.6" +"Motor vehicle theft . . . . .. .. . .... .","65.6","16.0","49.6","53.9","13.3","40.7","11.7","2.7","8.9" +"Arson .. . . . .. . ... .... .... .... .","9.8","4.3","5.5","8.1","3.7","4.4","1.7","0.6","1.1" +"Other assaults .. . . . . .. . ... . ... ..","1,061.3","175.3","886.1","785.4","115.4","670.0","276.0","59.9","216.1" +"Forgery and counterfeiting .. . . . . . ..","68.9","1.7","67.2","42.9","1.2","41.7","26.0","0.5","25.5" +"Fraud .... .. . . .. ... .... .... ....","173.7","5.1","168.5","98.4","3.3","95.0","75.3","1.8","73.5" +"Embezzlement . . .. . . . .. . ... . ....","14.6","–","14.1","7.2","–","6.9","7.4","–","7.2" +"Stolen property 1 . . . . . . .. . .. .. ...","84.3","15.1","69.2","66.7","12.2","54.5","17.6","2.8","14.7" +"Vandalism . . . . . . . .. .. .. .. .. ....","217.4","72.7","144.7","178.1","62.8","115.3","39.3","9.9","29.4" +"Weapons; carrying, possessing, etc. .","132.9","27.1","105.8","122.1","24.3","97.8","10.8","2.8","8.0" +"Prostitution and commercialized vice","56.9","1.1","55.8","17.3","–","17.1","39.6","0.8","38.7" +"Sex offenses 2 . . . . .. . . . .. .. .. . ..","61.5","10.7","50.7","56.1","9.6","46.5","5.4","1.1","4.3" +"Drug abuse violations . . . . . . . .. ...","1,333.0","136.6","1,196.4","1,084.3","115.2","969.1","248.7","21.4","227.3" +"Gambling .. . . . . .. ... . ... . ... ...","8.2","1.4","6.8","7.2","1.4","5.9","0.9","–","0.9" +"Offenses against the family and","","","","","","","","","" +"children . . . .. . . .. .. .. .. .. .. . ..","92.4","3.7","88.7","68.9","2.4","66.6","23.4","1.3","22.1" +"Driving under the influence . . . . . .. .","1,158.5","109.2","1,147.5","895.8","8.2","887.6","262.7","2.7","260.0" +"Liquor laws . . . . . . . .. .. .. .. .. .. .","48.2","90.2","368.0","326.8","55.4","271.4","131.4","34.7","96.6" +"Drunkenness . . .. . . . .. . ... . ... ..","488.1","11.4","476.8","406.8","8.5","398.3","81.3","2.9","78.4" +"Disorderly conduct . .. . . . . . .. .. .. .","529.5","136.1","393.3","387.1","90.8","296.2","142.4","45.3","97.1" +"Vagrancy . . . .. . . . ... .... .... ...","26.6","2.2","24.4","20.9","1.6","19.3","5.7","0.6","5.1" +"All other offenses (except traffic) . . ..","306.1","263.4","2,800.8","2,337.1","194.2","2,142.9","727.0","69.2","657.9" +"Suspicion . . . .. . . .. .. .. .. .. .. . ..","1.6","–","1.4","1.2","–","1.0","–","–","–" +"Curfew and loitering law violations ..","91.0","91.0","(X)","63.1","63.1","(X)","28.0","28.0","(X)" +"Runaways . . . . . . . .. .. .. .. .. ....","75.8","75.8","(X)","34.0","34.0","(X)","41.8","41.8","(X)" +"","– Represents zero. X Not applicable. 1 Buying, receiving, possessing stolen property. 2 Except forcible rape and prostitution.","","","","","","","","" diff --git a/docs/benchmark/stream/12s0324/12s0324-data-camelot-page-1-table-2.csv b/docs/benchmark/stream/12s0324/12s0324-data-camelot-page-1-table-2.csv new file mode 100755 index 0000000..b5b7f28 --- /dev/null +++ b/docs/benchmark/stream/12s0324/12s0324-data-camelot-page-1-table-2.csv @@ -0,0 +1,36 @@ +"","","","","American","" +"Offense charged","","","","Indian/Alaskan","Asian Pacific" +"","Total","White","Black","Native","Islander" +"Total . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .","10,690,561","7,389,208","3,027,153","150,544","123,656" +"Violent crime . . . . . . . . . . . . . . . . . . . . . . . . . . . .","456,965","268,346","177,766","5,608","5,245" +"Murder and nonnegligent manslaughter . .. ... .","9,739","4,741","4,801","100","97" +"Forcible rape . . . . . . . .. .. .. .. .... .. ...... .","16,362","10,644","5,319","169","230" +"Robbery . . . . .. . . . ... . ... . .... .... .... . . .","100,496","43,039","55,742","726","989" +"Aggravated assault . . . . . . . .. .. ...... .. ....","330,368","209,922","111,904","4,613","3,929" +"Property crime . . . . . . . . . . . . . . . . . . . . . . . . . . .","1,364,409","922,139","406,382","17,599","18,289" +"Burglary . . .. . . . .. . .... .... .... .... ... . . .","234,551","155,994","74,419","2,021","2,117" +"Larceny-theft . . . . . . . .. .. .. .. .... .. ...... .","1,056,473","719,983","306,625","14,646","15,219" +"Motor vehicle theft . . . . . .. ... . ... ..... ... ..","63,919","39,077","23,184","817","841" +"Arson .. . . .. .. .. ... .... .... .... .... . . . . .","9,466","7,085","2,154","115","112" +"Other assaults .. . . . . . ... . ... . ... ..... ... ..","1,032,502","672,865","332,435","15,127","12,075" +"Forgery and counterfeiting .. . . . . . ... ..... .. ..","67,054","44,730","21,251","345","728" +"Fraud ... . . . . .. .. .. .. .. .. .. .. .. .... . . . . . .","161,233","108,032","50,367","1,315","1,519" +"Embezzlement . . . .. . . . ... . ... . .... ... .....","13,960","9,208","4,429","75","248" +"Stolen property; buying, receiving, possessing .. .","82,714","51,953","29,357","662","742" +"Vandalism . . . . . . . .. .. .. .. .. .. .... .. ..... .","212,173","157,723","48,746","3,352","2,352" +"Weapons—carrying, possessing, etc. .. .. ... .. .","130,503","74,942","53,441","951","1,169" +"Prostitution and commercialized vice . ... .. .. ..","56,560","31,699","23,021","427","1,413" +"Sex offenses 1 . . . . . . . .. .. .. .. .... .. ...... .","60,175","44,240","14,347","715","873" +"Drug abuse violations . . . . . . . .. . ..... .. .....","1,301,629","845,974","437,623","8,588","9,444" +"Gambling . . . . .. . . . ... . ... . .. ... . ...... .. .","8,046","2,290","5,518","27","211" +"Offenses against the family and children ... .. .. .","87,232","58,068","26,850","1,690","624" +"Driving under the influence . . . . . . .. ... ...... .","1,105,401","954,444","121,594","14,903","14,460" +"Liquor laws . . . . . . . .. .. .. .. .. . ..... .. .....","444,087","373,189","50,431","14,876","5,591" +"Drunkenness . .. . . . . . ... . ... . ..... . .......","469,958","387,542","71,020","8,552","2,844" +"Disorderly conduct . . .. . . . . .. .. . ..... .. .....","515,689","326,563","176,169","8,783","4,174" +"Vagrancy . . .. .. . . .. ... .... .... .... .... . . .","26,347","14,581","11,031","543","192" +"All other offenses (except traffic) . .. .. .. ..... ..","2,929,217","1,937,221","911,670","43,880","36,446" +"Suspicion . . .. . . . .. .. .. .. .. .. .. ...... .. . . .","1,513","677","828","1","7" +"Curfew and loitering law violations . .. ... .. ....","89,578","54,439","33,207","872","1,060" +"Runaways . . . . . . . .. .. .. .. .. .. .... .. ..... .","73,616","48,343","19,670","1,653","3,950" +"1 Except forcible rape and prostitution.","","","","","" diff --git a/docs/benchmark/stream/12s0324/12s0324-data-tabula-0.csv.csv b/docs/benchmark/stream/12s0324/12s0324-data-tabula-0.csv.csv new file mode 100755 index 0000000..3de14fe --- /dev/null +++ b/docs/benchmark/stream/12s0324/12s0324-data-tabula-0.csv.csv @@ -0,0 +1,37 @@ +"",,Total,,,Male,,,Female, +Offense charged,,Under 18,18 years,,Under 18,18 years,,Under 18,18 years +"",Total,years,and over,Total,years,and over,Total,years,and over +Total . . . . . . . . . . . . . . . . . . . . . . . . .,"11,062 .6","1,540 .0","9,522 .6","8,263 .3","1,071 .6","7,191 .7","2,799 .2",468 .3,"2,330 .9" +Violent crime . . . . . . . . . . . . . . . . . .,467 .9,69 .1,398 .8,380 .2,56 .5,323 .7,87 .7,12 .6,75 .2 +Murder and nonnegligent,,,,,,,,, +manslaughter . . . . . . . . . . . . . . . . .,10.0,0.9,9.1,9.0,0.9,8.1,1.1,–,1.0 +Forcible rape . . . . . . . . . . . . . . . . . .,17.5,2.6,14.9,17.2,2.5,14.7,–,–,– +Robbery . . . . . . . . . . . . . . . . . . . . . .,102.1,25.5,76.6,90.0,22.9,67.1,12.1,2.5,9.5 +Aggravated assault . . . . . . . . . . . . .,338.4,40.1,298.3,264.0,30.2,233.8,74.4,9.9,64.5 +Property crime . . . . . . . . . . . . . . . . .,"1,396 .4",338 .7,"1,057 .7",875 .9,210 .8,665 .1,608 .2,127 .9,392 .6 +Burglary . . . . . . . . . . . . . . . . . . . . . .,240.9,60.3,180.6,205.0,53.4,151.7,35.9,6.9,29.0 +Larceny-theft . . . . . . . . . . . . . . . . . .,"1,080.1",258.1,822.0,608.8,140.5,468.3,471.3,117.6,353.6 +Motor vehicle theft . . . . . . . . . . . . . .,65.6,16.0,49.6,53.9,13.3,40.7,11.7,2.7,8.9 +Arson . . . . . . . . . . . . . . . . . . . . . . . .,9.8,4.3,5.5,8.1,3.7,4.4,1.7,0.6,1.1 +Other assaults . . . . . . . . . . . . . . . . . .,"1,061.3",175.3,886.1,785.4,115.4,670.0,276.0,59.9,216.1 +Forgery and counterfeiting . . . . . . . . .,68.9,1.7,67.2,42.9,1.2,41.7,26.0,0.5,25.5 +Fraud . . . . . . . . . . . . . . . . . . . . . . . . .,173.7,5.1,168.5,98.4,3.3,95.0,75.3,1.8,73.5 +Embezzlement . . . . . . . . . . . . . . . . . .,14.6,–,14.1,7.2,–,6.9,7.4,–,7.2 +Stolen property 1 . . . . . . . . . . . . . . . .,84.3,15.1,69.2,66.7,12.2,54.5,17.6,2.8,14.7 +Vandalism . . . . . . . . . . . . . . . . . . . . .,217.4,72.7,144.7,178.1,62.8,115.3,39.3,9.9,29.4 +"Weapons; carrying, possessing, etc. .",132.9,27.1,105.8,122.1,24.3,97.8,10.8,2.8,8.0 +Prostitution and commercialized vice,56.9,1.1,55.8,17.3,–,17.1,39.6,0.8,38.7 +Sex offenses 2 . . . . . . . . . . . . . . . . . .,61.5,10.7,50.7,56.1,9.6,46.5,5.4,1.1,4.3 +Drug abuse violations . . . . . . . . . . . .,"1,333.0",136.6,"1,196.4","1,084.3",115.2,969.1,248.7,21.4,227.3 +Gambling . . . . . . . . . . . . . . . . . . . . . .,8.2,1.4,6.8,7.2,1.4,5.9,0.9,–,0.9 +Offenses against the family and,,,,,,,,, +children . . . . . . . . . . . . . . . . . . . . . .,92.4,3.7,88.7,68.9,2.4,66.6,23.4,1.3,22.1 +Driving under the influence . . . . . . . .,"1,158.5",109.2,"1,147.5",895.8,8.2,887.6,262.7,2.7,260.0 +Liquor laws . . . . . . . . . . . . . . . . . . . .,48.2,90.2,368.0,326.8,55.4,271.4,131.4,34.7,96.6 +Drunkenness . . . . . . . . . . . . . . . . . . .,488.1,11.4,476.8,406.8,8.5,398.3,81.3,2.9,78.4 +Disorderly conduct . . . . . . . . . . . . . . .,529.5,136.1,393.3,387.1,90.8,296.2,142.4,45.3,97.1 +Vagrancy . . . . . . . . . . . . . . . . . . . . . .,26.6,2.2,24.4,20.9,1.6,19.3,5.7,0.6,5.1 +All other offenses (except traffic) . . . .,306.1,263.4,"2,800.8","2,337.1",194.2,"2,142.9",727.0,69.2,657.9 +Suspicion . . . . . . . . . . . . . . . . . . . . . .,1.6,–,1.4,1.2,–,1.0,–,–,– +Curfew and loitering law violations . .,91.0,91.0,(X),63.1,63.1,(X),28.0,28.0,(X) +Runaways . . . . . . . . . . . . . . . . . . . . .,75.8,75.8,(X),34.0,34.0,(X),41.8,41.8,(X) diff --git a/docs/benchmark/stream/12s0324/12s0324-data-tabula-1.csv.csv b/docs/benchmark/stream/12s0324/12s0324-data-tabula-1.csv.csv new file mode 100755 index 0000000..f51232e --- /dev/null +++ b/docs/benchmark/stream/12s0324/12s0324-data-tabula-1.csv.csv @@ -0,0 +1,35 @@ +"",,,,American, +Offense charged,,,,Indian/Alaskan,Asian Pacific +"",Total,White,Black,Native,Islander +Total . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .,"10,690,561","7,389,208","3,027,153","150,544","123,656" +Violent crime . . . . . . . . . . . . . . . . . . . . . . . . . . . .,"456,965","268,346","177,766","5,608","5,245" +Murder and nonnegligent manslaughter . . . . . . .,"9,739","4,741","4,801",100,97 +Forcible rape . . . . . . . . . . . . . . . . . . . . . . . . . . . .,"16,362","10,644","5,319",169,230 +Robbery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .,"100,496","43,039","55,742",726,989 +Aggravated assault . . . . . . . . . . . . . . . . . . . . . . .,"330,368","209,922","111,904","4,613","3,929" +Property crime . . . . . . . . . . . . . . . . . . . . . . . . . . .,"1,364,409","922,139","406,382","17,599","18,289" +Burglary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .,"234,551","155,994","74,419","2,021","2,117" +Larceny-theft . . . . . . . . . . . . . . . . . . . . . . . . . . . .,"1,056,473","719,983","306,625","14,646","15,219" +Motor vehicle theft . . . . . . . . . . . . . . . . . . . . . . . .,"63,919","39,077","23,184",817,841 +Arson . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .,"9,466","7,085","2,154",115,112 +Other assaults . . . . . . . . . . . . . . . . . . . . . . . . . . . .,"1,032,502","672,865","332,435","15,127","12,075" +Forgery and counterfeiting . . . . . . . . . . . . . . . . . . .,"67,054","44,730","21,251",345,728 +Fraud . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .,"161,233","108,032","50,367","1,315","1,519" +Embezzlement . . . . . . . . . . . . . . . . . . . . . . . . . . . .,"13,960","9,208","4,429",75,248 +"Stolen property; buying, receiving, possessing . . .","82,714","51,953","29,357",662,742 +Vandalism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .,"212,173","157,723","48,746","3,352","2,352" +"Weapons—carrying, possessing, etc. . . . . . . . . . .","130,503","74,942","53,441",951,"1,169" +Prostitution and commercialized vice . . . . . . . . . .,"56,560","31,699","23,021",427,"1,413" +Sex offenses 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . .,"60,175","44,240","14,347",715,873 +Drug abuse violations . . . . . . . . . . . . . . . . . . . . . .,"1,301,629","845,974","437,623","8,588","9,444" +Gambling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .,"8,046","2,290","5,518",27,211 +Offenses against the family and children . . . . . . . .,"87,232","58,068","26,850","1,690",624 +Driving under the influence . . . . . . . . . . . . . . . . . .,"1,105,401","954,444","121,594","14,903","14,460" +Liquor laws . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .,"444,087","373,189","50,431","14,876","5,591" +Drunkenness . . . . . . . . . . . . . . . . . . . . . . . . . . . . .,"469,958","387,542","71,020","8,552","2,844" +Disorderly conduct . . . . . . . . . . . . . . . . . . . . . . . . .,"515,689","326,563","176,169","8,783","4,174" +Vagrancy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .,"26,347","14,581","11,031",543,192 +All other offenses (except traffic) . . . . . . . . . . . . . .,"2,929,217","1,937,221","911,670","43,880","36,446" +Suspicion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .,"1,513",677,828,1,7 +Curfew and loitering law violations . . . . . . . . . . . .,"89,578","54,439","33,207",872,"1,060" +Runaways . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .,"73,616","48,343","19,670","1,653","3,950" diff --git a/docs/benchmark/stream/12s0324/12s0324.pdf b/docs/benchmark/stream/12s0324/12s0324.pdf new file mode 100755 index 0000000..c192726 Binary files /dev/null and b/docs/benchmark/stream/12s0324/12s0324.pdf differ diff --git a/docs/benchmark/stream/birdisland/birdisland-data-camelot-page-1-table-1.csv b/docs/benchmark/stream/birdisland/birdisland-data-camelot-page-1-table-1.csv new file mode 100755 index 0000000..a05fcaa --- /dev/null +++ b/docs/benchmark/stream/birdisland/birdisland-data-camelot-page-1-table-1.csv @@ -0,0 +1,35 @@ +"","","","","","SCN","Seed","Yield","Moisture","Lodgingg","g","Stand","","Gross" +"Company/Brandpy","","Product/Brand†","Technol.†","Mat.","Resist.","Trmt.†","Bu/A","%","%","","(x 1000)(",")","Income" +"KrugerKruger","","K2-1901K2 1901","RR2YRR2Y","1.91.9","RR","Ac,PVAc,PV","56.456.4","7.67.6","00","","126.3126.3","","$846$846" +"StineStine","","19RA02 §19RA02 §","RR2YRR2Y","1 91.9","RR","CMBCMB","55.355.3","7 67.6","00","","120 0120.0","","$830$830" +"WensmanWensman","","W 3190NR2W 3190NR2","RR2YRR2Y","1 91.9","RR","AcAc","54 554.5","7 67.6","00","","119 5119.5","","$818$818" +"H ftHefty","","H17Y12H17Y12","RR2YRR2Y","1 71.7","MRMR","II","53 753.7","7 77.7","00","","124 4124.4","","$806$806" +"Dyna-Gro","","S15RY53","RR2Y","1.5","R","Ac","53.6","7.7","0","","126.8","","$804" +"LG SeedsLG Seeds","","C2050R2C2050R2","RR2YRR2Y","2.12.1","RR","AcAc","53.653.6","7.77.7","00","","123.9123.9","","$804$804" +"Titan ProTitan Pro","","19M4219M42","RR2YRR2Y","1.91.9","RR","CMBCMB","53.653.6","7.77.7","00","","121.0121.0","","$804$804" +"StineStine","","19RA02 (2) §19RA02 (2) §","RR2YRR2Y","1 91.9","RR","CMBCMB","53 453.4","7 77.7","00","","123 9123.9","","$801$801" +"AsgrowAsgrow","","AG1832 §AG1832 §","RR2YRR2Y","1 81.8","MRMR","Ac PVAc,PV","52 952.9","7 77.7","00","","122 0122.0","","$794$794" +"Prairie Brandiid","","PB-1566R2662","RR2Y2","1.5","R","CMB","52.8","7.7","0","","122.9","","$792$" +"Channel","","1901R2","RR2Y","1.9","R","Ac,PV,","52.8","7.6","0","","123.4","","$791$" +"Titan ProTitan Pro","","20M120M1","RR2YRR2Y","2.02.0","RR","AmAm","52.552.5","7.57.5","00","","124.4124.4","","$788$788" +"KrugerKruger","","K2-2002K2-2002","RR2YRR2Y","2 02.0","RR","Ac PVAc,PV","52 452.4","7 97.9","00","","125 4125.4","","$786$786" +"ChannelChannel","","1700R21700R2","RR2YRR2Y","1 71.7","RR","Ac PVAc,PV","52 352.3","7 97.9","00","","123 9123.9","","$784$784" +"H ftHefty","","H16Y11H16Y11","RR2YRR2Y","1 61.6","MRMR","II","51 451.4","7 67.6","00","","123 9123.9","","$771$771" +"Anderson","","162R2Y","RR2Y","1.6","R","None","51.3","7.5","0","","119.5","","$770" +"Titan ProTitan Pro","","15M2215M22","RR2YRR2Y","1.51.5","RR","CMBCMB","51.351.3","7.87.8","00","","125.4125.4","","$769$769" +"DairylandDairyland","","DSR-1710R2YDSR-1710R2Y","RR2YRR2Y","1 71.7","RR","CMBCMB","51 351.3","7 77.7","00","","122 0122.0","","$769$769" +"HeftyHefty","","H20R3H20R3","RR2YRR2Y","2 02.0","MRMR","II","50 550.5","8 28.2","00","","121 0121.0","","$757$757" +"PPrairie BrandiiBd","","PB 1743R2PB-1743R2","RR2YRR2Y","1 71.7","RR","CMBCMB","50 250.2","7 77.7","00","","125 8125.8","","$752$752" +"Gold Country","","1741","RR2Y","1.7","R","Ac","50.1","7.8","0","","123.9","","$751" +"Trelaye ay","","20RR4303","RR2Y","2.00","R","Ac,Exc,","49.99 9","7.66","00","","127.88","","$749$9" +"HeftyHefty","","H14R3H14R3","RR2YRR2Y","1.41.4","MRMR","II","49.749.7","7.77.7","00","","122.9122.9","","$746$746" +"Prairie BrandPrairie Brand","","PB-2099NRR2PB-2099NRR2","RR2YRR2Y","2 02.0","RR","CMBCMB","49 649.6","7 87.8","00","","126 3126.3","","$743$743" +"WensmanWensman","","W 3174NR2W 3174NR2","RR2YRR2Y","1 71.7","RR","AcAc","49 349.3","7 67.6","00","","122 5122.5","","$740$740" +"KKruger","","K2 1602K2-1602","RR2YRR2Y","1 61.6","R","Ac,PV","48.78","7.66","00","","125.412","","$731$31" +"NK Brand","","S18-C2 §§","RR2Y","1.8","R","CMB","48.7","7.7","0","","126.8","","$731$" +"KrugerKruger","","K2-1902K2 1902","RR2YRR2Y","1.91.9","RR","Ac,PVAc,PV","48.748.7","7.57.5","00","","124.4124.4","","$730$730" +"Prairie BrandPrairie Brand","","PB-1823R2PB-1823R2","RR2YRR2Y","1 81.8","RR","NoneNone","48 548.5","7 67.6","00","","121 0121.0","","$727$727" +"Gold CountryGold Country","","15411541","RR2YRR2Y","1 51.5","RR","AcAc","48 448.4","7 67.6","00","","110 4110.4","","$726$726" +"","","","","","","Test Average =","47 647.6","7 77.7","00","","122 9122.9","","$713$713" +"","","","","","","LSD (0.10) =","5.7","0.3","ns","","37.8","","566.4" +"","F.I.R.S.T. Managerg","","","","","C.V. =","8.8","2.9","","","56.4","","846.2" diff --git a/docs/benchmark/stream/birdisland/birdisland-data-tabula.csv b/docs/benchmark/stream/birdisland/birdisland-data-tabula.csv new file mode 100755 index 0000000..1791ee9 --- /dev/null +++ b/docs/benchmark/stream/birdisland/birdisland-data-tabula.csv @@ -0,0 +1,30 @@ +Kruger,K2-1901,RR2Y,1.9,R,"Ac,PV",56.4,7.6,0,126.3,$846 +Stine,19RA02 §,RR2Y,1.9,R,CMB,55.3,7.6,0,120.0,$830 +Wensman,W 3190NR2,RR2Y,1.9,R,Ac,54.5,7.6,0,119.5,$818 +Hefty,H17Y12,RR2Y,1.7,MR,I,53.7,7.7,0,124.4,$806 +Dyna-Gro,S15RY53,RR2Y,1.5,R,Ac,53.6,7.7,0,126.8,$804 +LG Seeds,C2050R2,RR2Y,2.1,R,Ac,53.6,7.7,0,123.9,$804 +Titan Pro,19M42,RR2Y,1.9,R,CMB,53.6,7.7,0,121.0,$804 +Stine,19RA02 (2) §,RR2Y,1.9,R,CMB,53.4,7.7,0,123.9,$801 +Asgrow,AG1832 §,RR2Y,1.8,MR,"Ac,PV",52.9,7.7,0,122.0,$794 +Prairie Brand,PB-1566R2,RR2Y,1.5,R,CMB,52.8,7.7,0,122.9,$792 +Channel,1901R2,RR2Y,1.9,R,"Ac,PV",52.8,7.6,0,123.4,$791 +Titan Pro,20M1,RR2Y,2.0,R,Am,52.5,7.5,0,124.4,$788 +Kruger,K2-2002,RR2Y,2.0,R,"Ac,PV",52.4,7.9,0,125.4,$786 +Channel,1700R2,RR2Y,1.7,R,"Ac,PV",52.3,7.9,0,123.9,$784 +Hefty,H16Y11,RR2Y,1.6,MR,I,51.4,7.6,0,123.9,$771 +Anderson,162R2Y,RR2Y,1.6,R,None,51.3,7.5,0,119.5,$770 +Titan Pro,15M22,RR2Y,1.5,R,CMB,51.3,7.8,0,125.4,$769 +Dairyland,DSR-1710R2Y,RR2Y,1.7,R,CMB,51.3,7.7,0,122.0,$769 +Hefty,H20R3,RR2Y,2.0,MR,I,50.5,8.2,0,121.0,$757 +Prairie Brand,PB-1743R2,RR2Y,1.7,R,CMB,50.2,7.7,0,125.8,$752 +Gold Country,1741,RR2Y,1.7,R,Ac,50.1,7.8,0,123.9,$751 +Trelay,20RR43,RR2Y,2.0,R,"Ac,Ex",49.9,7.6,0,127.8,$749 +Hefty,H14R3,RR2Y,1.4,MR,I,49.7,7.7,0,122.9,$746 +Prairie Brand,PB-2099NRR2,RR2Y,2.0,R,CMB,49.6,7.8,0,126.3,$743 +Wensman,W 3174NR2,RR2Y,1.7,R,Ac,49.3,7.6,0,122.5,$740 +Kruger,K2-1602,RR2Y,1.6,R,"Ac,PV",48.7,7.6,0,125.4,$731 +NK Brand,S18-C2 §,RR2Y,1.8,R,CMB,48.7,7.7,0,126.8,$731 +Kruger,K2-1902,RR2Y,1.9,R,"Ac,PV",48.7,7.5,0,124.4,$730 +Prairie Brand,PB-1823R2,RR2Y,1.8,R,None,48.5,7.6,0,121.0,$727 +Gold Country,1541,RR2Y,1.5,R,Ac,48.4,7.6,0,110.4,$726 diff --git a/docs/benchmark/stream/birdisland/birdisland.pdf b/docs/benchmark/stream/birdisland/birdisland.pdf new file mode 100755 index 0000000..1501158 Binary files /dev/null and b/docs/benchmark/stream/birdisland/birdisland.pdf differ diff --git a/docs/benchmark/stream/budget/budget-data-camelot-page-1-table-1.csv b/docs/benchmark/stream/budget/budget-data-camelot-page-1-table-1.csv new file mode 100755 index 0000000..7b49956 --- /dev/null +++ b/docs/benchmark/stream/budget/budget-data-camelot-page-1-table-1.csv @@ -0,0 +1,46 @@ +"","Budget Provisions, net of receipts and recoveries, for the year 2014-2015 are as under.","","","","","","","" +"","Further details are on pages indicated in the last column of this Statement","","","","","","","" +"Budget 2014-2015","","","","","","","","" +"","","","","","","","(In crores of Rupees)","" +"","","Plan","","","Non-Plan","","Total Plan &","" +"","","","","","","","Non-Plan","" +"Ministry/Demand","Revenue","Capital","Total","Revenue","Capital","Total","","Page No." +"MINISTRY OF AGRICULTURE","28130.48","67.52","28198.00","2863.09","1.85","2864.94","31062.94","" +"1. Department of Agriculture and Cooperation","22260.55","48.45","22309.00","342.51","0.74","343.25","22652.25","1-10" +"2. Department of Agricultural Research and Education","3715.00","...","3715.00","2429.39","...","2429.39","6144.39","11-13" +"3. Department of Animal Husbandry, Dairying and Fisheries","2154.93","19.07","2174.00","91.19","1.11","92.30","2266.30","14-19" +"DEPARTMENT OF ATOMIC ENERGY","1779.00","4101.00","5880.00","3710.84","855.75","4566.59","10446.59","" +"4. Atomic Energy","1483.00","3427.00","4910.00","2971.25","855.75","3827.00","8737.00","20-25" +"5. Nuclear Power Schemes","296.00","674.00","970.00","739.59","...","739.59","1709.59","26-27" +"MINISTRY OF CHEMICALS AND FERTILISERS","360.83","153.17","514.00","73104.46","0.09","73104.55","73618.55","" +"6. Department of Chemicals and Petrochemicals","171.49","35.51","207.00","63.67","0.01","63.68","270.68","28-30" +"7. Department of Fertilisers","12.34","87.66","100.00","72999.96","0.04","73000.00","73100.00","31-33" +"8. Department of Pharmaceuticals","177.00","30.00","207.00","40.83","0.04","40.87","247.87","34-36" +"MINISTRY OF CIVIL AVIATION","179.90","6540.10","6720.00","657.98","...","657.98","7377.98","" +"9. Ministry of Civil Aviation","179.90","6540.10","6720.00","657.98","...","657.98","7377.98","37-39" +"MINISTRY OF COAL","550.00","...","550.00","50.00","...","50.00","600.00","" +"10. Ministry of Coal","550.00","...","550.00","50.00","...","50.00","600.00","40-41" +"MINISTRY OF COMMERCE AND INDUSTRY","3515.75","410.25","3926.00","3873.02","...","3873.02","7799.02","" +"11. Department of Commerce","1921.50","304.50","2226.00","3628.00","...","3628.00","5854.00","42-47" +"12. Department of Industrial Policy and Promotion","1594.25","105.75","1700.00","245.02","...","245.02","1945.02","48-52" +"MINISTRY OF COMMUNICATIONS AND INFORMATION TECHNOLOGY","7704.70","4410.30","12115.00","13953.92","101.00","14054.92","26169.92","" +"13. Department of Posts","286.70","513.30","800.00","6907.76","5.00","6912.76","7712.76","53-54" +"14. Department of Telecommunications","3798.00","3702.00","7500.00","6932.06","96.00","7028.06","14528.06","55-58" +"15. Department of Electronics and Information Technology","3620.00","195.00","3815.00","114.10","...","114.10","3929.10","59-63" +"MINISTRY OF CONSUMER AFFAIRS, FOOD AND PUBLIC DISTRIBUTION","344.45","205.55","550.00","115402.63","...","115402.63","115952.63","" +"16. Department of Consumer Affairs","192.70","27.30","220.00","75.79","...","75.79","295.79","64-66" +"17. Department of Food and Public Distribution","151.75","178.25","330.00","115326.84","...","115326.84","115656.84","67-70" +"MINISTRY OF CORPORATE AFFAIRS","22.76","1.24","24.00","208.75","22.50","231.25","255.25","" +"18. Ministry of Corporate Affairs","22.76","1.24","24.00","208.75","22.50","231.25","255.25","71-72" +"MINISTRY OF CULTURE","1767.00","68.00","1835.00","676.00","...","676.00","2511.00","" +"19. Ministry of Culture","1767.00","68.00","1835.00","676.00","...","676.00","2511.00","73-78" +"MINISTRY OF DEFENCE","...","...","...","188994.20","96208.67","285202.87","285202.87","" +"20. Ministry of Defence","...","...","...","3582.15","1620.72","5202.87","5202.87","79-80" +"21. Defence Pensions","...","...","...","51000.00","...","51000.00","51000.00","81-81" +"22. Defence Services-Army","...","...","...","92669.32","...","92669.32","92669.32","82-82" +"23. Defence Services-Navy","...","...","...","13975.79","...","13975.79","13975.79","83-83" +"24. Defence Services-Air Force","...","...","...","20506.84","...","20506.84","20506.84","84-84" +"25. Defence Ordnance Factories","...","...","...","1275.43","...","1275.43","1275.43","85-85" +"26. Defence Services – Research and Development","...","...","...","5984.67","...","5984.67","5984.67","86-86" +"27. Capital Outlay on Defence Services","...","...","...","...","94587.95","94587.95","94587.95","87-88" +"","","","","","","","SBE Summary of Contents","" diff --git a/docs/benchmark/stream/budget/budget-data-tabula.csv b/docs/benchmark/stream/budget/budget-data-tabula.csv new file mode 100755 index 0000000..4385259 --- /dev/null +++ b/docs/benchmark/stream/budget/budget-data-tabula.csv @@ -0,0 +1,43 @@ +"",Budget 2014-2015,,,,,,,(In crores of Rupees) +"",,,Plan,,,Non-Plan,,Total Plan & +Ministry/Demand,,Revenue,Capital,Total,Revenue,Capital,Total,Non-Plan Page No. +"",,,,,,,, +"",,,,,,,, +MINISTRY OF AGRICULTURE,,28130.48,67.52,28198.00,2863.09,1.85,2864.94,31062.94 +1.Department of Agriculture and Cooperation,,22260.55,48.45,22309.00,342.51,0.74,343.25,22652.25 1-10 +2.Department of Agricultural Research and Education,,3715.00,...,3715.00,2429.39,...,2429.39,6144.39 11-13 +"3.Department of Animal Husbandry, Dairying and Fisheries",,2154.93,19.07,2174.00,91.19,1.11,92.30,2266.30 14-19 +DEPARTMENT OF ATOMIC ENERGY,,1779.00,4101.00,5880.00,3710.84,855.75,4566.59,10446.59 +4.Atomic Energy,,1483.00,3427.00,4910.00,2971.25,855.75,3827.00,8737.00 20-25 +5.Nuclear Power Schemes,,296.00,674.00,970.00,739.59,...,739.59,1709.59 26-27 +MINISTRY OF CHEMICALS AND FERTILISERS,,360.83,153.17,514.00,73104.46,0.09,73104.55,73618.55 +6.Department of Chemicals and Petrochemicals,,171.49,35.51,207.00,63.67,0.01,63.68,270.68 28-30 +7.Department of Fertilisers,,12.34,87.66,100.00,72999.96,0.04,73000.00,73100.00 31-33 +8.Department of Pharmaceuticals,,177.00,30.00,207.00,40.83,0.04,40.87,247.87 34-36 +MINISTRY OF CIVIL AVIATION,,179.90,6540.10,6720.00,657.98,...,657.98,7377.98 +9.Ministry of Civil Aviation,,179.90,6540.10,6720.00,657.98,...,657.98,7377.98 37-39 +MINISTRY OF COAL,,550.00,...,550.00,50.00,...,50.00,600.00 +10.Ministry of Coal,,550.00,...,550.00,50.00,...,50.00,600.00 40-41 +MINISTRY OF COMMERCE AND INDUSTRY,,3515.75,410.25,3926.00,3873.02,...,3873.02,7799.02 +11.Department of Commerce,,1921.50,304.50,2226.00,3628.00,...,3628.00,5854.00 42-47 +12.Department of Industrial Policy and Promotion,,1594.25,105.75,1700.00,245.02,...,245.02,1945.02 48-52 +MINISTRY OF COMMUNICATIONS AND INFORMATION TECHNOLOGY,,7704.70,4410.30,12115.00,13953.92,101.00,14054.92,26169.92 +13.Department of Posts,,286.70,513.30,800.00,6907.76,5.00,6912.76,7712.76 53-54 +14.Department of Telecommunications,,3798.00,3702.00,7500.00,6932.06,96.00,7028.06,14528.06 55-58 +15.Department of Electronics and Information Technology,,3620.00,195.00,3815.00,114.10,...,114.10,3929.10 59-63 +"MINISTRY OF CONSUMER AFFAIRS, FOOD AND PUBLIC DISTRIBUTION",,344.45,205.55,550.00,115402.63,...,115402.63,115952.63 +16.Department of Consumer Affairs,,192.70,27.30,220.00,75.79,...,75.79,295.79 64-66 +17.Department of Food and Public Distribution,,151.75,178.25,330.00,115326.84,...,115326.84,115656.84 67-70 +MINISTRY OF CORPORATE AFFAIRS,,22.76,1.24,24.00,208.75,22.50,231.25,255.25 +18.Ministry of Corporate Affairs,,22.76,1.24,24.00,208.75,22.50,231.25,255.25 71-72 +MINISTRY OF CULTURE,,1767.00,68.00,1835.00,676.00,...,676.00,2511.00 +19.Ministry of Culture,,1767.00,68.00,1835.00,676.00,...,676.00,2511.00 73-78 +MINISTRY OF DEFENCE,,...,...,...,188994.20,96208.67,285202.87,285202.87 +20.Ministry of Defence,,...,...,...,3582.15,1620.72,5202.87,5202.87 79-80 +21.Defence Pensions,,...,...,...,51000.00,...,51000.00,51000.00 81-81 +22.Defence Services-Army,,...,...,...,92669.32,...,92669.32,92669.32 82-82 +23.Defence Services-Navy,,...,...,...,13975.79,...,13975.79,13975.79 83-83 +24.Defence Services-Air Force,,...,...,...,20506.84,...,20506.84,20506.84 84-84 +25.Defence Ordnance Factories,,...,...,...,1275.43,...,1275.43,1275.43 85-85 +26.Defence Services – Research and Development,,...,...,...,5984.67,...,5984.67,5984.67 86-86 +27.Capital Outlay on Defence Services,,...,...,...,...,94587.95,94587.95,94587.95 87-88 diff --git a/docs/benchmark/stream/budget/budget.pdf b/docs/benchmark/stream/budget/budget.pdf new file mode 100755 index 0000000..9466e87 Binary files /dev/null and b/docs/benchmark/stream/budget/budget.pdf differ diff --git a/docs/benchmark/stream/district_health/district_health-data-camelot-page-1-table-1.csv b/docs/benchmark/stream/district_health/district_health-data-camelot-page-1-table-1.csv new file mode 100755 index 0000000..9251cf6 --- /dev/null +++ b/docs/benchmark/stream/district_health/district_health-data-camelot-page-1-table-1.csv @@ -0,0 +1,50 @@ +"Chandel- Key Indicators","","","","" +"","DLHS-4 (2012-13)","","DLHS-3 (2007-08)","" +"Indicators","TOTAL","RURAL","TOTAL","RURAL" +"Child feeding practices (based on last-born child in the reference period) (%)","","","","" +"Children age 0-5 months exclusively breastfed9 ..........................................................................","76.9","80.0","NA","NA" +"Children age 6-9 months receiving solid/semi-solid food and breast milk ....................................","78.6","75.0","85.9","89.3" +"Children age 12-23 months receiving breast feeding along with complementary feeding ...........","31.8","24.2","NA","NA" +"Children age 6-35 months exclusively breastfed for at least 6 months ........................................","4.7","3.4","30.0","27.7" +"Children under 3 years breastfed within one hour of birth ............................................................","42.9","46.5","50.6","52.9" +"Birth Weight (%) (age below 36 months)","","","","" +"Percentage of Children weighed at birth ......................................................................................","38.8","41.0","NA","NA" +"Percentage of Children with low birth weight (out of those who weighted) ( below 2.5 kg) .........","12.8","14.6","NA","NA" +"Awareness about Diarrhoea (%)","","","","" +"Awareness about ARI (%)","","","","" +"Women know about what to do when a child gets diarrhoea .....................................................","96.3","96.2","94.4","94.2" +"Women aware about danger signs of ARI10 .................................................................................","55.9","59.7","32.8","34.7" +"Treatment of childhood diseases (based on last two surviving children born during the","","","","" +"reference period) (%)","","","","" +"Prevalence of diarrhoea in last 2 weeks for under 5 years old children .......................................","1.6","1.3","6.5","7.0" +"Children with diarrhoea in the last 2 weeks and received ORS11 .................................................","100.0","100.0","54.8","53.3" +"Children with diarrhoea in the last 2 weeks and sought advice/treatment ...................................","100.0","50.0","72.9","73.3" +"Prevalence of ARI in last 2 weeks for under 5 years old children ............................................","4.3","3.9","3.9","4.2" +"Children with acute respiratory infection or fever in last 2 weeks and sought advice/treatment","37.5","33.3","69.8","68.0" +"Children with diarrhoea in the last 2 weeks given Zinc along with ORS ......................................","66.6","50.0","NA","NA" +"Awareness of RTI/STI and HIV/AIDS (%)","","","","" +"Women who have heard of RTI/STI .............................................................................................","55.8","57.1","34.8","38.2" +"Women who have heard of HIV/AIDS ..........................................................................................","98.9","99.0","98.3","98.1" +"Women who have any symptoms of RTI/STI ..............................................................................","13.9","13.5","15.6","16.1" +"Women who know the place to go for testing of HIV/AIDS12 .......................................................","59.9","57.1","48.6","46.3" +"Women underwent test for detecting HIV/AIDS12 ........................................................................","37.3","36.8","14.1","12.3" +"Utilization of Government Health Services (%)","","","","" +"Antenatal care ..............................................................................................................................","69.7","66.7","79.0","81.0" +"Treatment for pregnancy complications .......................................................................................","57.1","59.3","88.0","87.8" +"Treatment for post-delivery complications ...................................................................................","33.3","33.3","68.4","68.4" +"Treatment for vaginal discharge ...................................................................................................","20.0","25.0","73.9","71.4" +"Treatment for children with diarrhoea13 ........................................................................................","50.0","100.0","NA","NA" +"Treatment for children with ARI13 .................................................................................................","NA","NA","NA","NA" +"Birth Registration (%)","","","","" +"Children below age 5 years having birth registration done ..........................................................","40.6","44.3","NA","NA" +"Children below age 5 years who received birth certificate (out of those registered) ....................","65.9","63.6","NA","NA" +"Personal Habits (age 15 years and above) (%)","","","","" +"Men who use any kind of smokeless tobacco .............................................................................","74.6","74.2","NA","NA" +"Women who use any kind of smokeless tobacco ........................................................................","59.5","58.9","NA","NA" +"Men who smoke ...........................................................................................................................","56.0","56.4","NA","NA" +"Women who smoke ......................................................................................................................","18.4","18.0","NA","NA" +"Men who consume alcohol ...........................................................................................................","58.4","58.2","NA","NA" +"Women who consume alcohol .....................................................................................................","10.9","9.3","NA","NA" +"9 Children Who were given nothing but breast milk till the survey date 10Acute Respiratory Infections 11Oral Rehydration Solutions/Salts.12Based on","","","","" +"the women who have heard of HIV/AIDS.13 Last two weeks","","","","" +"3","","","","" diff --git a/docs/benchmark/stream/district_health/district_health-data-tabula.csv b/docs/benchmark/stream/district_health/district_health-data-tabula.csv new file mode 100755 index 0000000..bd7fd72 --- /dev/null +++ b/docs/benchmark/stream/district_health/district_health-data-tabula.csv @@ -0,0 +1,48 @@ +"",DLHS-4 (2012-13) DLHS-3 (2007-08) +Indicators,TOTAL RURAL TOTAL RURAL +Child feeding practices (based on last-born child in the reference period) (%), +Children age 0-5 months exclusively breastfed9 ..........................................................................,76.9 80.0 NA NA +Children age 6-9 months receiving solid/semi-solid food and breast milk ....................................,78.6 75.0 85.9 89.3 +Children age 12-23 months receiving breast feeding along with complementary feeding ...........,31.8 24.2 NA NA +Children age 6-35 months exclusively breastfed for at least 6 months ........................................,4.7 3.4 30.0 27.7 +Children under 3 years breastfed within one hour of birth ............................................................,42.9 46.5 50.6 52.9 +Birth Weight (%) (age below 36 months), +Percentage of Children weighed at birth ......................................................................................,38.8 41.0 NA NA +Percentage of Children with low birth weight (out of those who weighted) ( below 2.5 kg) .........,12.8 14.6 NA NA +Awareness about Diarrhoea (%), +Women know about what to do when a child gets diarrhoea .....................................................,96.3 96.2 94.4 94.2 +Awareness about ARI (%), +Women aware about danger signs of ARI10 .................................................................................,55.9 59.7 32.8 34.7 +Treatment of childhood diseases (based on last two surviving children born during the, +reference period) (%), +Prevalence of diarrhoea in last 2 weeks for under 5 years old children .......................................,1.6 1.3 6.5 7.0 +Children with diarrhoea in the last 2 weeks and received ORS11 .................................................,100.0 100.0 54.8 53.3 +Children with diarrhoea in the last 2 weeks and sought advice/treatment ...................................,100.0 50.0 72.9 73.3 +Prevalence of ARI in last 2 weeks for under 5 years old children ............................................,4.3 3.9 3.9 4.2 +Children with acute respiratory infection or fever in last 2 weeks and sought advice/treatment,37.5 33.3 69.8 68.0 +Children with diarrhoea in the last 2 weeks given Zinc along with ORS ......................................,66.6 50.0 NA NA +Awareness of RTI/STI and HIV/AIDS (%), +Women who have heard of RTI/STI .............................................................................................,55.8 57.1 34.8 38.2 +Women who have heard of HIV/AIDS ..........................................................................................,98.9 99.0 98.3 98.1 +Women who have any symptoms of RTI/STI ..............................................................................,13.9 13.5 15.6 16.1 +Women who know the place to go for testing of HIV/AIDS12 .......................................................,59.9 57.1 48.6 46.3 +Women underwent test for detecting HIV/AIDS12 ........................................................................,37.3 36.8 14.1 12.3 +Utilization of Government Health Services (%), +Antenatal care ..............................................................................................................................,69.7 66.7 79.0 81.0 +Treatment for pregnancy complications .......................................................................................,57.1 59.3 88.0 87.8 +Treatment for post-delivery complications ...................................................................................,33.3 33.3 68.4 68.4 +Treatment for vaginal discharge ...................................................................................................,20.0 25.0 73.9 71.4 +Treatment for children with diarrhoea13 ........................................................................................,50.0 100.0 NA NA +Treatment for children with ARI13 .................................................................................................,NA NA NA NA +Birth Registration (%), +Children below age 5 years having birth registration done ..........................................................,40.6 44.3 NA NA +Children below age 5 years who received birth certificate (out of those registered) ....................,65.9 63.6 NA NA +Personal Habits (age 15 years and above) (%), +Men who use any kind of smokeless tobacco .............................................................................,74.6 74.2 NA NA +Women who use any kind of smokeless tobacco ........................................................................,59.5 58.9 NA NA +Men who smoke ...........................................................................................................................,56.0 56.4 NA NA +Women who smoke ......................................................................................................................,18.4 18.0 NA NA +Men who consume alcohol ...........................................................................................................,58.4 58.2 NA NA +Women who consume alcohol .....................................................................................................,10.9 9.3 NA NA +9 Children Who were given nothing but breast milk till the survey date 10Acute Respiratory Infections 11Oral Rehydration Solutions/Salts.12Based on, +the women who have heard of HIV/AIDS.13 Last two weeks, diff --git a/docs/benchmark/stream/district_health/district_health.pdf b/docs/benchmark/stream/district_health/district_health.pdf new file mode 100755 index 0000000..3ab299b Binary files /dev/null and b/docs/benchmark/stream/district_health/district_health.pdf differ diff --git a/docs/benchmark/stream/health/health-data-camelot-page-1-table-1.csv b/docs/benchmark/stream/health/health-data-camelot-page-1-table-1.csv new file mode 100755 index 0000000..311468e --- /dev/null +++ b/docs/benchmark/stream/health/health-data-camelot-page-1-table-1.csv @@ -0,0 +1,31 @@ +"","Table: 5 Public Health Outlay 2012-13 (Budget Estimates) (Rs. in 000)","","","","","","" +"States-A","Revenue","","Capital","","Total","Others(1)","Total" +"","","","","","Revenue &","","" +"","Medical &","Family","Medical &","Family","","","" +"","","","","","Capital","","" +"","Public","Welfare","Public","Welfare","","","" +"","Health","","Health","","","","" +"Andhra Pradesh","47,824,589","9,967,837","1,275,000","15,000","59,082,426","14,898,243","73,980,669" +"Arunachal Pradesh","2,241,609","107,549","23,000","0","2,372,158","86,336","2,458,494" +"Assam","14,874,821","2,554,197","161,600","0","17,590,618","4,408,505","21,999,123" +"Bihar","21,016,708","4,332,141","5,329,000","0","30,677,849","2,251,571","32,929,420" +"Chhattisgarh","11,427,311","1,415,660","2,366,592","0","15,209,563","311,163","15,520,726" +"Delhi","28,084,780","411,700","4,550,000","0","33,046,480","5,000","33,051,480" +"Goa","4,055,567","110,000","330,053","0","4,495,620","12,560","4,508,180" +"Gujarat","26,328,400","6,922,900","12,664,000","42,000","45,957,300","455,860","46,413,160" +"Haryana","15,156,681","1,333,527","40,100","0","16,530,308","1,222,698","17,753,006" +"Himachal Pradesh","8,647,229","1,331,529","580,800","0","10,559,558","725,315","11,284,873" +"Jammu & Kashmir","14,411,984","270,840","3,188,550","0","17,871,374","166,229","18,037,603" +"Jharkhand","8,185,079","3,008,077","3,525,558","0","14,718,714","745,139","15,463,853" +"Karnataka","34,939,843","4,317,801","3,669,700","0","42,927,344","631,088","43,558,432" +"Kerala","27,923,965","3,985,473","929,503","0","32,838,941","334,640","33,173,581" +"Madhya Pradesh","28,459,540","4,072,016","3,432,711","0","35,964,267","472,139","36,436,406" +"Maharashtra","55,011,100","6,680,721","5,038,576","0","66,730,397","313,762","67,044,159" +"Manipur","2,494,600","187,700","897,400","0","3,579,700","0","3,579,700" +"Meghalaya","2,894,093","342,893","705,500","5,000","3,947,486","24,128","3,971,614" +"Mizoram","1,743,501","84,185","10,250","0","1,837,936","17,060","1,854,996" +"Nagaland","2,368,724","204,329","226,400","0","2,799,453","783,054","3,582,507" +"Odisha","14,317,179","2,552,292","1,107,250","0","17,976,721","451,438","18,428,159" +"Puducherry","4,191,757","52,249","192,400","0","4,436,406","2,173","4,438,579" +"Punjab","19,775,485","2,208,343","2,470,882","0","24,454,710","1,436,522","25,891,232" +"","Health Sector Financing by Centre and States/UTs in India [2009-10 to 2012-13](Revised) P a g e |23","","","","","","" diff --git a/docs/benchmark/stream/health/health-data-tabula.csv b/docs/benchmark/stream/health/health-data-tabula.csv new file mode 100755 index 0000000..29fc687 --- /dev/null +++ b/docs/benchmark/stream/health/health-data-tabula.csv @@ -0,0 +1,28 @@ +Table: 5,Public Health Outlay 2012-13 (Budget Estimates)(Rs. in 000), +States-A,Revenue Capital Total Others(1),Total +"",Medical & Family Medical & Family Revenue &, +"",Public Welfare Public Welfare Capital, +"",Health Health, +Andhra Pradesh,"47,824,589 9,967,837 1,275,000 15,000 59,082,426 14,898,243","73,980,669" +Arunachal Pradesh,"2,241,609 107,549 23,000 0 2,372,158 86,336","2,458,494" +Assam,"14,874,821 2,554,197 161,600 0 17,590,618 4,408,505","21,999,123" +Bihar,"21,016,708 4,332,141 5,329,000 0 30,677,849 2,251,571","32,929,420" +Chhattisgarh,"11,427,311 1,415,660 2,366,592 0 15,209,563 311,163","15,520,726" +Delhi,"28,084,780 411,700 4,550,000 0 33,046,480 5,000","33,051,480" +Goa,"4,055,567 110,000 330,053 0 4,495,620 12,560","4,508,180" +Gujarat,"26,328,400 6,922,900 12,664,000 42,000 45,957,300 455,860","46,413,160" +Haryana,"15,156,681 1,333,527 40,100 0 16,530,308 1,222,698","17,753,006" +Himachal Pradesh,"8,647,229 1,331,529 580,800 0 10,559,558 725,315","11,284,873" +Jammu & Kashmir,"14,411,984 270,840 3,188,550 0 17,871,374 166,229","18,037,603" +Jharkhand,"8,185,079 3,008,077 3,525,558 0 14,718,714 745,139","15,463,853" +Karnataka,"34,939,843 4,317,801 3,669,700 0 42,927,344 631,088","43,558,432" +Kerala,"27,923,965 3,985,473 929,503 0 32,838,941 334,640","33,173,581" +Madhya Pradesh,"28,459,540 4,072,016 3,432,711 0 35,964,267 472,139","36,436,406" +Maharashtra,"55,011,100 6,680,721 5,038,576 0 66,730,397 313,762","67,044,159" +Manipur,"2,494,600 187,700 897,400 0 3,579,700 0","3,579,700" +Meghalaya,"2,894,093 342,893 705,500 5,000 3,947,486 24,128","3,971,614" +Mizoram,"1,743,501 84,185 10,250 0 1,837,936 17,060","1,854,996" +Nagaland,"2,368,724 204,329 226,400 0 2,799,453 783,054","3,582,507" +Odisha,"14,317,179 2,552,292 1,107,250 0 17,976,721 451,438","18,428,159" +Puducherry,"4,191,757 52,249 192,400 0 4,436,406 2,173","4,438,579" +Punjab,"19,775,485 2,208,343 2,470,882 0 24,454,710 1,436,522","25,891,232" diff --git a/docs/benchmark/stream/health/health.pdf b/docs/benchmark/stream/health/health.pdf new file mode 100755 index 0000000..b9247ab Binary files /dev/null and b/docs/benchmark/stream/health/health.pdf differ diff --git a/docs/benchmark/stream/m27/m27-data-camelot-page-1-table-1.csv b/docs/benchmark/stream/m27/m27-data-camelot-page-1-table-1.csv new file mode 100755 index 0000000..2d2539e --- /dev/null +++ b/docs/benchmark/stream/m27/m27-data-camelot-page-1-table-1.csv @@ -0,0 +1,47 @@ +"FEB","RUAR","Y 2014 M27 (BUS)","","ALPHABETIC LISTING BY T","YPE","","","","ABLPDM27" +"","","","","OF ACTIVE LICENSES","","","","","3/19/2014" +"","","","","OKLAHOMA ABLE COMMIS","SION","","","","" +"LICENSE","","","","PREMISE","","","","","" +"NUMBER","TYPE","DBA NAME","LICENSEE NAME","ADDRESS","CITY","ST","ZIP","PHONE NUMBER","EXPIRES" +"648765","AAA","ALLEGIANT AIR","ALLEGIANT AIR LLC","7100 TERMINAL DRIVE","OKLAHOMA CITY","OK","73159","-","2014/12/03" +"","","","","7777 EAST APACHE","","","","","" +"648766","AAA","ALLEGIANT AIR","ALLEGIANT AIR LLC","STREET","TULSA","OK","74115","-","2014/12/16" +"82030","AAA","AMERICAN AIRLINES","AMERICAN AIRLINES INC","7100 TERMINAL DRIVE","OKLAHOMA CITY","OK","73159","(405) 680-3701","2014/09/14" +"509462","AAA","AMERICAN AIRLINES","AMERICAN AIRLINES INC","7777 EAST APACHE DRIVE","TULSA","OK","74115","(918) 831-6302","2014/08/19" +"","","","AMERICAN EAGLE","","","","","","" +"509609","AAA","AMERICAN EAGLE","AIRLINES INC","7100 TERMINAL DRIVE","OKLAHOMA CITY","OK","73159","(405) 680-3701","2014/08/19" +"","","","AMERICAN EAGLE","","","","","","" +"402986","AAA","AMERICAN EAGLE","AIRLINES INC","7777 EAST APACHE DRIVE","TULSA","OK","74115","(859) 767-3747","2014/10/22" +"","","","","WILL ROGERS AIRPORT","","","","","" +"79145","AAA","DELTA AIR LINES","DELTA AIR LINES INC","BOX 59975","OKLAHOMA CITY","OK","73159","(404) 773-9745","2014/05/11" +"600941","AAA","ENDEAVOR AIR","ENDEAVOR AIR INC","7100 TERMINAL DRIVE","OKLAHOMA CITY","OK","73159","(901) 348-4100","2015/03/26" +"","","","","7100 TERMINAL DRIVE","","","","","" +"478482","AAA","EXPRESSJET AIRLINES","EXPRESSJET AIRLINES INC","WILL ROGERS AIRPORT","OKLAHOMA CITY","OK","73159","(832) 353-1201","2014/05/08" +"505981","AAA","SKYWEST AIRLINES","SKYWEST INC","7100 TERMINAL DRIVE","OKLAHOMA CITY","OK","73159","(405) 634-3000","2014/05/28" +"429754","AAA","SOUTHWEST AIRLINES","SOUTHWEST AIRLINES CO","7100 TERMINAL DRIVE","OKLAHOMA CITY","OK","73159","(405) 682-4183","2015/02/15" +"","","TULSA INTERNATIONAL","","","","","","","" +"429755","AAA","AIRPORT","SOUTHWEST AIRLINES CO","7777 EAST APACHE DRIVE","TULSA","OK","74115","(918) 834-4495","2015/02/16" +"415051","AAA","UNITED AIRLINES","UNITED AIRLINES INC","7777 EAST APACHE DRIVE","TULSA","OK","74115","(872) 825-8309","2014/05/12" +"106719","AAA","UNITED AIRLINES","UNITED AIRLINES INC","WILL ROGERS AIRPORT","OKLAHOMA CITY","OK","73159","(872) 825-8309","2014/04/11" +"","","A SENSU JAPANESE","","7123 SOUTH 92ND EAST","","","","","" +"625422","BAW","RESTAURANT","INFORMAL PARTNERSHIP","AVENUE SUITE J","TULSA","OK","74133","(918) 252-0333","2015/02/14" +"","","ADAMO'S ROUTE 66","","2132 WEST GARY","","","","","" +"464828","BAW","ITALIAN VILLA","TADJ INC","BOULEVARD","CLINTON","OK","73601","(580) 323-5900","2015/02/11" +"","","","","12215 NORTH","","","","","" +"184066","BAW","AJANTA","CABAB N' CURRY INC","PENNSYLVANIA","OKLAHOMA CITY","OK","73120","(405) 752-5283","2014/07/27" +"","","","SAYRE LODGING","","","","","","" +"547693","BAW","AMERICINN OF SAYRE","ENTERPRISES LLC","2405 SOUTH EL CAMINO","SAYRE","OK","73662","(580) 928-2700","2014/09/08" +"","","ANDOLINI'S PIZZERIA &","","12140 EAST 96TH STREET","","","","","" +"428377","BAW","ITALIAN RESTAURANT","ANDOLINI'S LLC","NORTH #106","OWASSO","OK","74055","(918) 272-9325","2015/02/10" +"","","ASAHI JAPANESE","","","","","","","" +"446957","BAW","RESTAURANT","JIN CORPORATION","7831 EAST 71ST STREET","TULSA","OK","74133","(918) 307-9151","2014/12/22" +"","","","SMOKEHOUSE","","","","","","" +"632501","BAW","BACK DOOR BARBECUE","ASSOCIATES INC","315 NORTHWEST 23RD","OKLAHOMA CITY","OK","73103","-","2014/08/01" +"598515","BAW","BAMBOO THAI BISTRO","BAMBOO THAI BISTRO INC","5079 SOUTH YALE AVENUE","TULSA","OK","74135","(918) 828-0740","2015/03/11" +"","","BANDANA RED'S","","","","","","","" +"618693","BAW","STEAKHOUSE","BRADSHAW, STEVE_LEN","37808 OLD HIGHWAY 270","SHAWNEE","OK","74804","-","2014/08/20" +"","","","","1522 WEST LINDSEY","","","","","" +"632575","BAW","BASHU LEGENDS","HYH HE CHUANG LLC","STREET","NORMAN","OK","73069","-","2014/07/21" +"","","","DEEP FORK HOLDINGS","","","","","","" +"543149","BAW","BEDLAM BAR-B-Q","LLC","610 NORTHEAST 50TH","OKLAHOMA CITY","OK","73105","(405) 528-7427","2015/02/23" +"","","","","Page 1 of 151","","","","","" diff --git a/docs/benchmark/stream/m27/m27-data-tabula-0.csv b/docs/benchmark/stream/m27/m27-data-tabula-0.csv new file mode 100755 index 0000000..58648c8 --- /dev/null +++ b/docs/benchmark/stream/m27/m27-data-tabula-0.csv @@ -0,0 +1,41 @@ +648765 AAA,ALLEGIANT AIR,ALLEGIANT AIR LLC7100 TERMINAL DRIVE,OKLAHOMA CITY,,OK,73159,-,2014/12/03 +"",,7777 EAST APACHE,,,,,, +648766 AAA,ALLEGIANT AIR,ALLEGIANT AIR LLCSTREET,TULSA,,OK,74115,-,2014/12/16 +82030 AAA,AMERICAN AIRLINES,AMERICAN AIRLINES INC7100 TERMINAL DRIVE,OKLAHOMA CITY,,OK,73159,(405) 680-3701,2014/09/14 +509462 AAA,AMERICAN AIRLINES,AMERICAN AIRLINES INC7777 EAST APACHE DRIVE,TULSA,,OK,74115,(918) 831-6302,2014/08/19 +"",,AMERICAN EAGLE,,,,,, +509609 AAA,AMERICAN EAGLE,AIRLINES INC7100 TERMINAL DRIVE,OKLAHOMA CITY,,OK,73159,(405) 680-3701,2014/08/19 +"",,AMERICAN EAGLE,,,,,, +402986 AAA,AMERICAN EAGLE,AIRLINES INC7777 EAST APACHE DRIVE,TULSA,,OK,74115,(859) 767-3747,2014/10/22 +"",,WILL ROGERS AIRPORT,,,,,, +79145 AAA,DELTA AIR LINES,DELTA AIR LINES INCBOX 59975,OKLAHOMA CITY,,OK,73159,(404) 773-9745,2014/05/11 +600941 AAA,ENDEAVOR AIR,ENDEAVOR AIR INC7100 TERMINAL DRIVE,OKLAHOMA CITY,,OK,73159,(901) 348-4100,2015/03/26 +"",,7100 TERMINAL DRIVE,,,,,, +478482 AAA,EXPRESSJET AIRLINES,EXPRESSJET AIRLINES INC WILL ROGERS AIRPORT,OKLAHOMA CITY,,OK,73159,(832) 353-1201,2014/05/08 +505981 AAA,SKYWEST AIRLINES,SKYWEST INC7100 TERMINAL DRIVE,OKLAHOMA CITY,,OK,73159,(405) 634-3000,2014/05/28 +429754 AAA,SOUTHWEST AIRLINES,SOUTHWEST AIRLINES CO 7100 TERMINAL DRIVE,OKLAHOMA CITY,,OK,73159,(405) 682-4183,2015/02/15 +"",TULSA INTERNATIONAL,,,,,,, +429755 AAA,AIRPORT,SOUTHWEST AIRLINES CO 7777 EAST APACHE DRIVE,TULSA,,OK,74115,(918) 834-4495,2015/02/16 +415051 AAA,UNITED AIRLINES,UNITED AIRLINES INC7777 EAST APACHE DRIVE,TULSA,,OK,74115,(872) 825-8309,2014/05/12 +106719 AAA,UNITED AIRLINES,UNITED AIRLINES INCWILL ROGERS AIRPORT,OKLAHOMA CITY,,OK,73159,(872) 825-8309,2014/04/11 +"",A SENSU JAPANESE,7123 SOUTH 92ND EAST,,,,,, +625422 BAW,RESTAURANT,INFORMAL PARTNERSHIPAVENUE SUITE J,TULSA,,OK,74133,(918) 252-0333,2015/02/14 +"",ADAMO'S ROUTE 66,2132 WEST GARY,,,,,, +464828 BAW,ITALIAN VILLA,TADJ INCBOULEVARD,CLINTON,,OK,73601,(580) 323-5900,2015/02/11 +"",,12215 NORTH,,,,,, +184066 BAW,AJANTA,CABAB N' CURRY INCPENNSYLVANIA,OKLAHOMA CITY,,OK,73120,(405) 752-5283,2014/07/27 +"",,SAYRE LODGING,,,,,, +547693 BAW,AMERICINN OF SAYRE,ENTERPRISES LLC2405 SOUTH EL CAMINO,SAYRE,,OK,73662,(580) 928-2700,2014/09/08 +"",ANDOLINI'S PIZZERIA &,12140 EAST 96TH STREET,,,,,, +428377 BAW,ITALIAN RESTAURANT,ANDOLINI'S LLCNORTH #106,OWASSO,,OK,74055,(918) 272-9325,2015/02/10 +"",ASAHI JAPANESE,,,,,,, +446957 BAW,RESTAURANT,JIN CORPORATION7831 EAST 71ST STREET,TULSA,,OK,74133,(918) 307-9151,2014/12/22 +"",,SMOKEHOUSE,,,,,, +632501 BAW,BACK DOOR BARBECUE,ASSOCIATES INC315 NORTHWEST 23RD,OKLAHOMA CITY,,OK,73103,-,2014/08/01 +598515 BAW,BAMBOO THAI BISTRO,BAMBOO THAI BISTRO INC 5079 SOUTH YALE AVENUE TULSA,,,OK,74135,(918) 828-0740,2015/03/11 +"",BANDANA RED'S,,,,,,, +618693 BAW,STEAKHOUSE,"BRADSHAW, STEVE_LEN37808 OLD HIGHWAY 270",SHAWNEE,,OK,74804,-,2014/08/20 +"",,1522 WEST LINDSEY,,,,,, +632575 BAW,BASHU LEGENDS,HYH HE CHUANG LLCSTREET,NORMAN,,OK,73069,-,2014/07/21 +"",,DEEP FORK HOLDINGS,,,,,, +543149 BAW,BEDLAM BAR-B-Q,LLC610 NORTHEAST 50TH,OKLAHOMA CITY,,OK,73105,(405) 528-7427,2015/02/23 diff --git a/docs/benchmark/stream/m27/m27.pdf b/docs/benchmark/stream/m27/m27.pdf new file mode 100755 index 0000000..cecd7b6 Binary files /dev/null and b/docs/benchmark/stream/m27/m27.pdf differ diff --git a/docs/benchmark/stream/mexican_towns/mexican_towns-data-camelot-page-1-table-1.csv b/docs/benchmark/stream/mexican_towns/mexican_towns-data-camelot-page-1-table-1.csv new file mode 100755 index 0000000..e2a6a45 --- /dev/null +++ b/docs/benchmark/stream/mexican_towns/mexican_towns-data-camelot-page-1-table-1.csv @@ -0,0 +1,44 @@ +"Clave","","Clave","","","Clave","" +"","Nombre Entidad","","","Nombre Municipio","","Nombre Localidad" +"Entidad","","Municipio","","","Localidad","" +"01","Aguascalientes","001","Aguascalientes","","0094","Granja Adelita" +"01","Aguascalientes","001","Aguascalientes","","0096","Agua Azul" +"01","Aguascalientes","001","Aguascalientes","","0100","Rancho Alegre" +"01","Aguascalientes","001","Aguascalientes","","0102","Los Arbolitos [Rancho]" +"01","Aguascalientes","001","Aguascalientes","","0104","Ardillas de Abajo (Las Ardillas)" +"01","Aguascalientes","001","Aguascalientes","","0106","Arellano" +"01","Aguascalientes","001","Aguascalientes","","0112","Bajío los Vázquez" +"01","Aguascalientes","001","Aguascalientes","","0113","Bajío de Montoro" +"01","Aguascalientes","001","Aguascalientes","","0114","Residencial San Nicolás [Baños la Cantera]" +"01","Aguascalientes","001","Aguascalientes","","0120","Buenavista de Peñuelas" +"01","Aguascalientes","001","Aguascalientes","","0121","Cabecita 3 Marías (Rancho Nuevo)" +"01","Aguascalientes","001","Aguascalientes","","0125","Cañada Grande de Cotorina" +"01","Aguascalientes","001","Aguascalientes","","0126","Cañada Honda [Estación]" +"01","Aguascalientes","001","Aguascalientes","","0127","Los Caños" +"01","Aguascalientes","001","Aguascalientes","","0128","El Cariñán" +"01","Aguascalientes","001","Aguascalientes","","0129","El Carmen [Granja]" +"01","Aguascalientes","001","Aguascalientes","","0135","El Cedazo (Cedazo de San Antonio)" +"01","Aguascalientes","001","Aguascalientes","","0138","Centro de Arriba (El Taray)" +"01","Aguascalientes","001","Aguascalientes","","0139","Cieneguilla (La Lumbrera)" +"01","Aguascalientes","001","Aguascalientes","","0141","Cobos" +"01","Aguascalientes","001","Aguascalientes","","0144","El Colorado (El Soyatal)" +"01","Aguascalientes","001","Aguascalientes","","0146","El Conejal" +"01","Aguascalientes","001","Aguascalientes","","0157","Cotorina de Abajo" +"01","Aguascalientes","001","Aguascalientes","","0162","Coyotes" +"01","Aguascalientes","001","Aguascalientes","","0166","La Huerta (La Cruz)" +"01","Aguascalientes","001","Aguascalientes","","0170","Cuauhtémoc (Las Palomas)" +"01","Aguascalientes","001","Aguascalientes","","0171","Los Cuervos (Los Ojos de Agua)" +"01","Aguascalientes","001","Aguascalientes","","0172","San José [Granja]" +"01","Aguascalientes","001","Aguascalientes","","0176","La Chiripa" +"01","Aguascalientes","001","Aguascalientes","","0182","Dolores" +"01","Aguascalientes","001","Aguascalientes","","0183","Los Dolores" +"01","Aguascalientes","001","Aguascalientes","","0190","El Duraznillo" +"01","Aguascalientes","001","Aguascalientes","","0191","Los Durón" +"01","Aguascalientes","001","Aguascalientes","","0197","La Escondida" +"01","Aguascalientes","001","Aguascalientes","","0201","Brande Vin [Bodegas]" +"01","Aguascalientes","001","Aguascalientes","","0207","Valle Redondo" +"01","Aguascalientes","001","Aguascalientes","","0209","La Fortuna" +"01","Aguascalientes","001","Aguascalientes","","0212","Lomas del Gachupín" +"01","Aguascalientes","001","Aguascalientes","","0213","El Carmen (Gallinas Güeras) [Rancho]" +"01","Aguascalientes","001","Aguascalientes","","0216","La Gloria" +"01","Aguascalientes","001","Aguascalientes","","0226","Hacienda Nueva" diff --git a/docs/benchmark/stream/mexican_towns/mexican_towns-data-tabula.csv b/docs/benchmark/stream/mexican_towns/mexican_towns-data-tabula.csv new file mode 100755 index 0000000..9fc5afb --- /dev/null +++ b/docs/benchmark/stream/mexican_towns/mexican_towns-data-tabula.csv @@ -0,0 +1,41 @@ +01,Aguascalientes,001,Aguascalientes,0094,Granja Adelita +01,Aguascalientes,001,Aguascalientes,0096,Agua Azul +01,Aguascalientes,001,Aguascalientes,0100,Rancho Alegre +01,Aguascalientes,001,Aguascalientes,0102,Los Arbolitos [Rancho] +01,Aguascalientes,001,Aguascalientes,0104,Ardillas de Abajo (Las Ardillas) +01,Aguascalientes,001,Aguascalientes,0106,Arellano +01,Aguascalientes,001,Aguascalientes,0112,Bajío los Vázquez +01,Aguascalientes,001,Aguascalientes,0113,Bajío de Montoro +01,Aguascalientes,001,Aguascalientes,0114,Residencial San Nicolás [Baños la Cantera] +01,Aguascalientes,001,Aguascalientes,0120,Buenavista de Peñuelas +01,Aguascalientes,001,Aguascalientes,0121,Cabecita 3 Marías (Rancho Nuevo) +01,Aguascalientes,001,Aguascalientes,0125,Cañada Grande de Cotorina +01,Aguascalientes,001,Aguascalientes,0126,Cañada Honda [Estación] +01,Aguascalientes,001,Aguascalientes,0127,Los Caños +01,Aguascalientes,001,Aguascalientes,0128,El Cariñán +01,Aguascalientes,001,Aguascalientes,0129,El Carmen [Granja] +01,Aguascalientes,001,Aguascalientes,0135,El Cedazo (Cedazo de San Antonio) +01,Aguascalientes,001,Aguascalientes,0138,Centro de Arriba (El Taray) +01,Aguascalientes,001,Aguascalientes,0139,Cieneguilla (La Lumbrera) +01,Aguascalientes,001,Aguascalientes,0141,Cobos +01,Aguascalientes,001,Aguascalientes,0144,El Colorado (El Soyatal) +01,Aguascalientes,001,Aguascalientes,0146,El Conejal +01,Aguascalientes,001,Aguascalientes,0157,Cotorina de Abajo +01,Aguascalientes,001,Aguascalientes,0162,Coyotes +01,Aguascalientes,001,Aguascalientes,0166,La Huerta (La Cruz) +01,Aguascalientes,001,Aguascalientes,0170,Cuauhtémoc (Las Palomas) +01,Aguascalientes,001,Aguascalientes,0171,Los Cuervos (Los Ojos de Agua) +01,Aguascalientes,001,Aguascalientes,0172,San José [Granja] +01,Aguascalientes,001,Aguascalientes,0176,La Chiripa +01,Aguascalientes,001,Aguascalientes,0182,Dolores +01,Aguascalientes,001,Aguascalientes,0183,Los Dolores +01,Aguascalientes,001,Aguascalientes,0190,El Duraznillo +01,Aguascalientes,001,Aguascalientes,0191,Los Durón +01,Aguascalientes,001,Aguascalientes,0197,La Escondida +01,Aguascalientes,001,Aguascalientes,0201,Brande Vin [Bodegas] +01,Aguascalientes,001,Aguascalientes,0207,Valle Redondo +01,Aguascalientes,001,Aguascalientes,0209,La Fortuna +01,Aguascalientes,001,Aguascalientes,0212,Lomas del Gachupín +01,Aguascalientes,001,Aguascalientes,0213,El Carmen (Gallinas Güeras) [Rancho] +01,Aguascalientes,001,Aguascalientes,0216,La Gloria +01,Aguascalientes,001,Aguascalientes,0226,Hacienda Nueva diff --git a/docs/benchmark/stream/mexican_towns/mexican_towns.pdf b/docs/benchmark/stream/mexican_towns/mexican_towns.pdf new file mode 100755 index 0000000..46cd236 Binary files /dev/null and b/docs/benchmark/stream/mexican_towns/mexican_towns.pdf differ diff --git a/docs/benchmark/stream/missing_values/missing_values-data-camelot-page-1-table-1.csv b/docs/benchmark/stream/missing_values/missing_values-data-camelot-page-1-table-1.csv new file mode 100755 index 0000000..18e75d9 --- /dev/null +++ b/docs/benchmark/stream/missing_values/missing_values-data-camelot-page-1-table-1.csv @@ -0,0 +1,44 @@ +"Bhandara - Key Indicators","","","","" +"","DLHS-4 (2012-13)","","DLHS-3 (2007-08)","" +"Indicators","TOTAL","RURAL","TOTAL","RURAL" +"Reported Prevalence of Morbidity","","","","" +"Any Injury .....................................................................................................................................","1.9","2.1","","" +"Acute Illness .................................................................................................................................","4.5","5.6","","" +"Chronic Illness ..............................................................................................................................","5.1","4.1","","" +"Reported Prevalence of Chronic Illness during last one year (%)","","","","" +"Disease of respiratory system ......................................................................................................","11.7","15.0","","" +"Disease of cardiovascular system ................................................................................................","8.9","9.3","","" +"Persons suffering from tuberculosis .............................................................................................","2.2","1.5","","" +"Anaemia Status by Haemoglobin Level14 (%)","","","","" +"Children (6-59 months) having anaemia ......................................................................................","68.5","71.9","","" +"Children (6-59 months) having severe anaemia ..........................................................................","6.7","9.4","","" +"Children (6-9 Years) having anaemia - Male ................................................................................","67.1","71.4","","" +"Children (6-9 Years) having severe anaemia - Male ....................................................................","4.4","2.4","","" +"Children (6-9 Years) having anaemia - Female ...........................................................................","52.4","48.8","","" +"Children (6-9 Years) having severe anaemia - Female ................................................................","1.2","0.0","","" +"Children (6-14 years) having anaemia - Male .............................................................................","50.8","62.5","","" +"Children (6-14 years) having severe anaemia - Male ..................................................................","3.7","3.6","","" +"Children (6-14 years) having anaemia - Female .........................................................................","48.3","50.0","","" +"Children (6-14 years) having severe anaemia - Female ..............................................................","4.3","6.1","","" +"Children (10-19 Years15) having anaemia - Male .........................................................................","37.9","51.2","","" +"Children (10-19 Years15) having severe anaemia - Male .............................................................","3.5","4.0","","" +"Children (10-19 Years15) having anaemia - Female .....................................................................","46.6","52.1","","" +"Children (10-19 Years15) having severe anaemia - Female .........................................................","6.4","6.5","","" +"Adolescents (15-19 years) having anaemia ................................................................................","39.4","46.5","","" +"Adolescents (15-19 years) having severe anaemia .....................................................................","5.4","5.1","","" +"Pregnant women (15-49 aged) having anaemia ..........................................................................","48.8","51.5","","" +"Pregnant women (15-49 aged) having severe anaemia ..............................................................","7.1","8.8","","" +"Women (15-49 aged) having anaemia .........................................................................................","45.2","51.7","","" +"Women (15-49 aged) having severe anaemia .............................................................................","4.8","5.9","","" +"Persons (20 years and above) having anaemia ...........................................................................","37.8","42.1","","" +"Persons (20 years and above) having Severe anaemia ..............................................................","4.6","4.8","","" +"Blood Sugar Level (age 18 years and above) (%)","","","","" +"Blood Sugar Level >140 mg/dl (high) ...........................................................................................","12.9","11.1","","" +"Blood Sugar Level >160 mg/dl (very high) ...................................................................................","7.0","5.1","","" +"Hypertension (age 18 years and above) (%)","","","","" +"Above Normal Range (Systolic >140 mm of Hg & Diastolic >90 mm of Hg ) ..............................","23.8","22.8","","" +"Moderately High (Systolic >160 mm of Hg & Diastolic >100 mm of Hg ) .....................................","8.2","7.1","","" +"Very High (Systolic >180 mm of Hg & Diastolic >110 mm of Hg ) ...............................................","3.7","3.1","","" +"14 Any anaemia below 11g/dl, severe anaemia below 7g/dl. 15 Excluding age group 19 years","","","","" +"Chronic Illness :Any person with symptoms persisting for longer than one month is defined as suffering from chronic illness","","","","" +"4","","","","" diff --git a/docs/benchmark/stream/missing_values/missing_values-data-tabula.csv b/docs/benchmark/stream/missing_values/missing_values-data-tabula.csv new file mode 100755 index 0000000..826209d --- /dev/null +++ b/docs/benchmark/stream/missing_values/missing_values-data-tabula.csv @@ -0,0 +1,42 @@ +"",DLHS-4 (2012-13),DLHS-3 (2007-08) +Indicators,TOTAL RURAL,TOTAL RURAL +Reported Prevalence of Morbidity,, +Any Injury .....................................................................................................................................,1.9 2.1, +Acute Illness .................................................................................................................................,4.5 5.6, +Chronic Illness ..............................................................................................................................,5.1 4.1, +Reported Prevalence of Chronic Illness during last one year (%),, +Disease of respiratory system ......................................................................................................,11.7 15.0, +Disease of cardiovascular system ................................................................................................,8.9 9.3, +Persons suffering from tuberculosis .............................................................................................,2.2 1.5, +Anaemia Status by Haemoglobin Level14 (%),, +Children (6-59 months) having anaemia ......................................................................................,68.5 71.9, +Children (6-59 months) having severe anaemia ..........................................................................,6.7 9.4, +Children (6-9 Years) having anaemia - Male ................................................................................,67.1 71.4, +Children (6-9 Years) having severe anaemia - Male ....................................................................,4.4 2.4, +Children (6-9 Years) having anaemia - Female ...........................................................................,52.4 48.8, +Children (6-9 Years) having severe anaemia - Female ................................................................,1.2 0.0, +Children (6-14 years) having anaemia - Male .............................................................................,50.8 62.5, +Children (6-14 years) having severe anaemia - Male ..................................................................,3.7 3.6, +Children (6-14 years) having anaemia - Female .........................................................................,48.3 50.0, +Children (6-14 years) having severe anaemia - Female ..............................................................,4.3 6.1, +Children (10-19 Years15) having anaemia - Male .........................................................................,37.9 51.2, +Children (10-19 Years15) having severe anaemia - Male .............................................................,3.5 4.0, +Children (10-19 Years15) having anaemia - Female .....................................................................,46.6 52.1, +Children (10-19 Years15) having severe anaemia - Female .........................................................,6.4 6.5, +Adolescents (15-19 years) having anaemia ................................................................................,39.4 46.5, +Adolescents (15-19 years) having severe anaemia .....................................................................,5.4 5.1, +Pregnant women (15-49 aged) having anaemia ..........................................................................,48.8 51.5, +Pregnant women (15-49 aged) having severe anaemia ..............................................................,7.1 8.8, +Women (15-49 aged) having anaemia .........................................................................................,45.2 51.7, +Women (15-49 aged) having severe anaemia .............................................................................,4.8 5.9, +Persons (20 years and above) having anaemia ...........................................................................,37.8 42.1, +Persons (20 years and above) having Severe anaemia ..............................................................,4.6 4.8, +Blood Sugar Level (age 18 years and above) (%),, +Blood Sugar Level >140 mg/dl (high) ...........................................................................................,12.9 11.1, +Blood Sugar Level >160 mg/dl (very high) ...................................................................................,7.0 5.1, +Hypertension (age 18 years and above) (%),, +Above Normal Range (Systolic >140 mm of Hg & Diastolic >90 mm of Hg ) ..............................,23.8 22.8, +Moderately High (Systolic >160 mm of Hg & Diastolic >100 mm of Hg ) .....................................,8.2 7.1, +Very High (Systolic >180 mm of Hg & Diastolic >110 mm of Hg ) ...............................................,3.7 3.1, +"14 Any anaemia below 11g/dl, severe anaemia below 7g/dl. 15 Excluding age group 19 years",, +Chronic Illness :Any person with symptoms persisting for longer than one month is defined as suffering from chronic illness,, diff --git a/docs/benchmark/stream/missing_values/missing_values.pdf b/docs/benchmark/stream/missing_values/missing_values.pdf new file mode 100755 index 0000000..90b620f Binary files /dev/null and b/docs/benchmark/stream/missing_values/missing_values.pdf differ diff --git a/docs/benchmark/stream/population_growth/population_growth-data-camelot-page-1-table-1.csv b/docs/benchmark/stream/population_growth/population_growth-data-camelot-page-1-table-1.csv new file mode 100755 index 0000000..b642e18 --- /dev/null +++ b/docs/benchmark/stream/population_growth/population_growth-data-camelot-page-1-table-1.csv @@ -0,0 +1,42 @@ +"","TABLE 4: STATE-WISE DECADAL GROWTH RATE OF POPULATION","","","","","" +"","","","","","","(Per cent)" +"States/Union Territories","1951-1961","1961-1971","1971-1981","1981-1991","1991-2001","2001-2011" +"Andaman and Nicobar Islands","105.19","81.17","63.93","48.70","26.90","6.68" +"Andhra Pradesh","15.65","20.90","23.10","24.20","14.59","11.10" +"Arunachal Pradesh","-","38.91","35.15","36.83","27.00","25.92" +"Assam","34.98","34.95","23.36","24.24","18.92","16.93" +"Bihar","19.79","20.91","24.16","23.38","28.62","25.07" +"Chandigarh","394.13","114.59","75.55","42.16","40.28","17.10" +"Chhattisgarh","22.77","27.12","20.39","25.73","18.27","22.59" +"Dadra and Nagar Haveli","39.56","27.96","39.78","33.57","59.22","55.50" +"Daman and Diu","-24.56","70.85","26.07","28.62","55.73","53.54" +"NCT of Delhi","52.44","52.93","53.00","51.45","47.02","20.96" +"Goa","7.77","34.77","26.74","16.08","15.21","8.17" +"Gujarat","26.88","29.39","27.67","21.19","22.66","19.17" +"Haryana","33.79","32.22","28.75","27.41","28.43","19.90" +"Himachal Pradesh","17.87","23.04","23.71","20.79","17.54","12.81" +"Jammu and Kashmir","9.44","29.65","29.69","30.89","29.43","23.71" +"Jharkhand","19.69","22.58","23.79","24.03","23.36","22.34" +"Karnataka","21.57","24.22","26.75","21.12","17.51","15.67" +"Kerala","24.76","26.29","19.24","14.32","9.43","4.86" +"Lakshadweep","14.61","31.95","26.53","28.47","17.30","6.23" +"Madhya Pradesh","24.73","29.28","27.16","27.24","24.26","20.30" +"Maharashtra","23.60","27.45","24.54","25.73","22.73","15.99" +"Manipur","35.04","37.53","32.46","29.29","24.86","18.65" +"Meghalaya","27.03","31.50","32.04","32.86","30.65","27.82" +"Mizoram","35.61","24.93","48.55","39.70","28.82","22.78" +"Nagaland","14.07","39.88","50.05","56.08","64.53","-0.47" +"Odisha","19.82","25.05","20.17","20.06","16.25","13.97" +"Puducherry","16.34","27.81","28.15","33.64","20.62","27.72" +"Punjab","21.56","21.70","23.89","20.81","20.10","13.73" +"Rajasthan","26.20","27.83","32.97","28.44","28.41","21.44" +"Sikkim","17.76","29.38","50.77","28.47","33.06","12.36" +"Tamil Nadu","11.85","22.30","17.50","15.39","11.72","15.60" +"Tripura","78.71","36.28","31.92","34.30","16.03","14.75" +"Uttar Pradesh","16.38","19.54","25.39","25.61","25.85","20.09" +"Uttarakhand","22.57","24.42","27.45","23.13","20.41","19.17" +"West Bengal","32.80","26.87","23.17","24.73","17.77","13.93" +"ALL INDIA","21.51","24.80","24.66","23.87","21.54","17.64" +"‘-’: Not Available.","","","","","","" +"Source : Registrar General of India, Ministry of Home Affairs, Government of India.","","","","","","" +"","","6","","","","" diff --git a/docs/benchmark/stream/population_growth/population_growth-data-tabula.csv b/docs/benchmark/stream/population_growth/population_growth-data-tabula.csv new file mode 100755 index 0000000..defbfad --- /dev/null +++ b/docs/benchmark/stream/population_growth/population_growth-data-tabula.csv @@ -0,0 +1,37 @@ +States/Union Territories,1951-1961,1961-1971,1971-1981,1981-1991,1991-2001,2001-2011 +Andaman and Nicobar Islands,105.19,81.17,63.93,48.70,26.90,6.68 +Andhra Pradesh,15.65,20.90,23.10,24.20,14.59,11.10 +Arunachal Pradesh,-,38.91,35.15,36.83,27.00,25.92 +Assam,34.98,34.95,23.36,24.24,18.92,16.93 +Bihar,19.79,20.91,24.16,23.38,28.62,25.07 +Chandigarh,394.13,114.59,75.55,42.16,40.28,17.10 +Chhattisgarh,22.77,27.12,20.39,25.73,18.27,22.59 +Dadra and Nagar Haveli,39.56,27.96,39.78,33.57,59.22,55.50 +Daman and Diu,-24.56,70.85,26.07,28.62,55.73,53.54 +NCT of Delhi,52.44,52.93,53.00,51.45,47.02,20.96 +Goa,7.77,34.77,26.74,16.08,15.21,8.17 +Gujarat,26.88,29.39,27.67,21.19,22.66,19.17 +Haryana,33.79,32.22,28.75,27.41,28.43,19.90 +Himachal Pradesh,17.87,23.04,23.71,20.79,17.54,12.81 +Jammu and Kashmir,9.44,29.65,29.69,30.89,29.43,23.71 +Jharkhand,19.69,22.58,23.79,24.03,23.36,22.34 +Karnataka,21.57,24.22,26.75,21.12,17.51,15.67 +Kerala,24.76,26.29,19.24,14.32,9.43,4.86 +Lakshadweep,14.61,31.95,26.53,28.47,17.30,6.23 +Madhya Pradesh,24.73,29.28,27.16,27.24,24.26,20.30 +Maharashtra,23.60,27.45,24.54,25.73,22.73,15.99 +Manipur,35.04,37.53,32.46,29.29,24.86,18.65 +Meghalaya,27.03,31.50,32.04,32.86,30.65,27.82 +Mizoram,35.61,24.93,48.55,39.70,28.82,22.78 +Nagaland,14.07,39.88,50.05,56.08,64.53,-0.47 +Odisha,19.82,25.05,20.17,20.06,16.25,13.97 +Puducherry,16.34,27.81,28.15,33.64,20.62,27.72 +Punjab,21.56,21.70,23.89,20.81,20.10,13.73 +Rajasthan,26.20,27.83,32.97,28.44,28.41,21.44 +Sikkim,17.76,29.38,50.77,28.47,33.06,12.36 +Tamil Nadu,11.85,22.30,17.50,15.39,11.72,15.60 +Tripura,78.71,36.28,31.92,34.30,16.03,14.75 +Uttar Pradesh,16.38,19.54,25.39,25.61,25.85,20.09 +Uttarakhand,22.57,24.42,27.45,23.13,20.41,19.17 +West Bengal,32.80,26.87,23.17,24.73,17.77,13.93 +ALL INDIA,21.51,24.80,24.66,23.87,21.54,17.64 diff --git a/docs/benchmark/stream/population_growth/population_growth.pdf b/docs/benchmark/stream/population_growth/population_growth.pdf new file mode 100755 index 0000000..41a166e Binary files /dev/null and b/docs/benchmark/stream/population_growth/population_growth.pdf differ diff --git a/docs/benchmark/stream/us-007/us-007-data-camelot-page-1-table-1.csv b/docs/benchmark/stream/us-007/us-007-data-camelot-page-1-table-1.csv new file mode 100755 index 0000000..13b1dbc --- /dev/null +++ b/docs/benchmark/stream/us-007/us-007-data-camelot-page-1-table-1.csv @@ -0,0 +1,11 @@ +"","One Withholding" +"Payroll Period","Allowance" +"Weekly","$71.15" +"Biweekly","142.31" +"Semimonthly","154.17" +"Monthly","308.33" +"Quarterly","925.00" +"Semiannually","1,850.00" +"Annually","3,700.00" +"Daily or Miscellaneous","14.23" +"(each day of the payroll period)","" diff --git a/docs/benchmark/stream/us-007/us-007-data-tabula.csv b/docs/benchmark/stream/us-007/us-007-data-tabula.csv new file mode 100755 index 0000000..d2d3cc5 --- /dev/null +++ b/docs/benchmark/stream/us-007/us-007-data-tabula.csv @@ -0,0 +1,11 @@ +"",One Withholding +Payroll Period,Allowance +Weekly,$ 71.15 +Biweekly,142.31 +Semimonthly,154.17 +Monthly,308.33 +Quarterly,925.00 +Semiannually,"1,850.00" +Annually,"3,700.00" +Daily or Miscellaneous,14.23 +"(each day of the payroll period)", diff --git a/docs/benchmark/stream/us-007/us-007.pdf b/docs/benchmark/stream/us-007/us-007.pdf new file mode 100755 index 0000000..45b3de3 Binary files /dev/null and b/docs/benchmark/stream/us-007/us-007.pdf differ diff --git a/docs/benchmark/twotables_2/twotables_2-data-tabula.csv b/docs/benchmark/twotables_2/twotables_2-data-tabula.csv deleted file mode 100755 index df7938d..0000000 --- a/docs/benchmark/twotables_2/twotables_2-data-tabula.csv +++ /dev/null @@ -1,32 +0,0 @@ -State -State -State,,Literacy Status,,,,,, -"",State,n,,,,,, -"",,,Illiterate,Write,1-4 std.,5-8 std.,9-12 std.,College -"",,,,,,,, -"",Kerala,2400,7.2,0.5,25.3,20.1,41.5,5.5 -"",Tamil Nadu,2400,21.4,2.3,8.8,35.5,25.8,6.2 -"",Karnataka,2399,37.4,2.8,12.5,18.3,23.1,5.8 -"",Andhra Pradesh,2400,54.0,1.7,8.4,13.2,18.8,3.9 -"",Maharashtra,2400,22.0,0.9,17.3,20.3,32.6,7.0 -"",Gujarat,2390,28.6,0.1,14.4,23.1,26.9,6.8 -"",Madhya Pradesh,2402,29.1,3.4,8.5,35.1,13.3,10.6 -"",Orissa,2405,33.2,1.0,10.4,25.7,21.2,8.5 -"",West Bengal,2293,41.7,4.4,13.2,17.1,21.2,2.4 -"",Uttar Pradesh,2400,35.3,2.1,4.5,23.3,27.1,7.6 -State -State -State,,Literacy Status,,,,,, -"",State,n,,,,,, -"",,,Illiterate,Write,1-4 std.,5-8 std.,9-12 std.,College -"",,,,,,,, -"",Kerala,2400,8.8,0.3,20.1,17.0,45.6,8.2 -"",Tamil Nadu,2400,29.9,1.5,8.5,33.1,22.3,4.8 -"",Karnataka,2399,47.9,2.5,10.2,18.8,18.4,2.3 -"",Andhra Pradesh,2400,66.4,0.7,6.8,12.9,11.4,1.8 -"",Maharashtra,2400,41.3,0.6,14.1,20.1,21.6,2.2 -"",Gujarat,2390,57.6,0.1,10.3,16.5,12.9,2.7 -"",Madhya Pradesh,2402,58.7,2.2,6.6,24.1,5.3,3.0 -"",Orissa,2405,50.0,0.9,8.1,21.9,15.1,4.0 -"",West Bengal,2293,49.1,4.8,11.2,16.8,17.1,1.1 -"",Uttar Pradesh,2400,67.3,2.0,3.1,17.2,7.7,2.7