embee/pyaccuwage
2
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

Record merging seems to work now that header offsets have been corrected.

Browse source 
There's an issue parsing p1220 on line 2570. Maybe making the parser ignore
full-width lines during parsing would fix the problem, if there's some
way to check the length of a row, only counting single-spaced words?
This commit is contained in:
Binh 2013-01-29 15:48:32 -06:00
parent 6e4a975cfb
commit e6e087ef38
2 changed files with 7 additions and 10 deletions
Download patch file Download diff file Expand all files Collapse all files

9
pyaccuwage/pdfextract.py
View file

@ -22,7 +22,6 @@ class PDFRecordFinder(object):
self.field_heading_exp = field_heading_exp self.field_heading_exp = field_heading_exp
def records(self): def records(self):
#headings = self.locate_heading_rows()
headings = self.locate_heading_rows_by_field() headings = self.locate_heading_rows_by_field()
for (start, end, name) in headings: for (start, end, name) in headings:
@ -55,7 +54,7 @@ class PDFRecordFinder(object):
position -= 1 position -= 1
name = ''.join(header).strip().decode('ascii','ignore') name = ''.join(header).strip().decode('ascii','ignore')
results.append((i, name)) results.append((i, name, position))
else: else:
# See if this row forces us to break from field reading. # See if this row forces us to break from field reading.
if re.search('Record\ Layout', row): if re.search('Record\ Layout', row):
@ -65,7 +64,7 @@ class PDFRecordFinder(object):
for (a, b) in zip(results, results[1:] + [(len(self.textrows), None)]): for (a, b) in zip(results, results[1:] + [(len(self.textrows), None)]):
end_pos = None end_pos = None
print a[0], record_break[0], b[0]-1 #print a[0], record_break[0], b[0]-1
while record_break and record_break[0] < a[0]: while record_break and record_break[0] < a[0]:
record_break = record_break[1:] record_break = record_break[1:]
@ -76,7 +75,7 @@ class PDFRecordFinder(object):
else: else:
end_pos = b[0]-1 end_pos = b[0]-1
merged.append( (a[0], end_pos, a[1]) ) merged.append( (a[0], end_pos-1, a[1]) )
return merged return merged
def locate_heading_rows(self): def locate_heading_rows(self):
@ -110,7 +109,6 @@ class PDFRecordFinder(object):
for r in row_iter: for r in row_iter:
row = self.extract_columns_from_row(r.decode('UTF-8')) row = self.extract_columns_from_row(r.decode('UTF-8'))
if not row: if not row:
cc.empty_row() cc.empty_row()
continue continue
@ -249,7 +247,6 @@ class ColumnCollector(object):
def adjust_columns(self, data): def adjust_columns(self, data):
adjusted_data = {} adjusted_data = {}
for col_id, value in data: for col_id, value in data:
if col_id in self.data.keys(): if col_id in self.data.keys():
adjusted_data[col_id] = value.strip() adjusted_data[col_id] = value.strip()

8
scripts/pyaccuwage-pdfparse
View file

@ -65,11 +65,11 @@ for rec in records:
# a lump of text and not necessarily a field entry. I assume # a lump of text and not necessarily a field entry. I assume
# this is cleaned out by the record builder class. # this is cleaned out by the record builder class.
print last_record_ends_at + 1, begins_at #print last_record_ends_at + 1, begins_at
#if last_record_ends_at + 1 != begins_at: if last_record_ends_at + 1 != begins_at:
sys.stdout.write("\nclass %s(object):\n" % re.sub('[^\w]','',rec[0])) sys.stdout.write("\nclass %s(object):\n" % re.sub('[^\w]','',rec[0].split(':')[-1]))
for field in builder.load(map(lambda x:x.tuple, rec[1][1:])): for field in builder.load(map(lambda x:x.tuple, rec[1][0:])):
sys.stdout.write('\t' + field + '\n') sys.stdout.write('\t' + field + '\n')
#print field #print field