From c8ba6bf55647136879833d0c0ada6865c82e33d9 Mon Sep 17 00:00:00 2001
From: Kragen Javier Sitaker <xentrac@special-circumstanc.es>
Date: Thu, 1 Jul 2021 03:02:57 -0300
Subject: [PATCH] Clean up pdftour code slightly
---
pdftour.py | 93 +++++++++++++++++++++++++++---------------------------
1 file changed, 47 insertions(+), 46 deletions(-)
diff --git a/pdftour.py b/pdftour.py
index b043658..676408e 100755
--- a/pdftour.py
+++ b/pdftour.py
@@ -120,7 +120,8 @@ clearly than a lower-level language would.
A pitfall I've run into a lot with the PEG parser in here is that if
you try to parse a Unicode string with a byte-string grammar or vice
-versa, you just get silent failures to parse.
+versa, you just get silent failures to parse. This is a specific case
+of the more general problem with silent failures.
Still unimplemented:
@@ -656,7 +657,7 @@ def ps_grammar():
arraydelim = Charset(set(b'[]'))
arraydelim.xform = lambda d: ('ad', d)
- tokens = Any(drop_ws(integer | op | name| dictdelim | arraydelim | parenstring | hexstring))
+ tokens = Any(drop_ws(integer | op | name | dictdelim | arraydelim | parenstring | hexstring))
root = Any(wschar) + tokens
root.xform = lambda d: d[1]
return root
@@ -664,49 +665,7 @@ def ps_grammar():
def tokenize_ps(ps):
return Parse(ps).do(0, ps_grammar())
-
-# XXX move these to the bottom
-def buildtables(tokens):
- "Given a sequence of tokens from a CMap, pull out the tables they represent."
- csranges = []
- bfchars = []
- bfranges = []
- sizes = {}
-
- def n_strings_back(n):
- for j in range(i-n, i):
- assert tokens[j][0] == 'str'
- yield tokens[j][1]
-
- for i, tok in enumerate(tokens):
- if tok[0] == 'op':
- op = tok[1]
- if op in [b'begincodespacerange', b'beginbfchar', b'beginbfrange']:
- assert tokens[i-1][0] == 'int'
- sizes[op] = tokens[i-1][1]
- elif op == b'endcodespacerange':
- csranges.extend(n_strings_back(2*sizes[b'begincodespacerange']))
- elif op == b'endbfchar':
- bfchars.extend(n_strings_back(2*sizes[b'beginbfchar']))
- elif op == b'endbfrange':
- bfranges.extend(n_strings_back(3*sizes[b'beginbfrange']))
-
- return csranges, bfchars, bfranges
-
-
-def csranges_to_grammar(csranges):
- "Compile the csranges from buildtables into a grammar we can use to tokenize strings."
- alternatives = []
- for i in range(0, len(csranges), 2):
- ranges = zip(csranges[i], csranges[i+1])
- cs = [Charset(set(c for c in range(startbyte, endbyte+1)))
- for startbyte, endbyte in ranges]
- prod = functools.reduce(operator.add, cs)
- prod.xform = bytes
- alternatives.append(prod)
-
- return Any(as_prod(alternatives))
-
+# PDF constructs not used in PS:
xref_header = Lit(b'xref') + eol + integer + b' ' + integer + eol # clause 7.5.4
# XXX This xform clearly indicates that I need to rethink how
@@ -777,6 +736,7 @@ class Pdf:
return plumb
+
class XrefSection:
def __init__(self, pdf, offset):
self.pdf = pdf
@@ -917,7 +877,7 @@ def read_pdf(filename):
return Pdf(fo.read())
-### Stuff for walking the PDF tree and pulling out CMaps ###
+### CMap handling, including stuff for walking the PDF tree to find CMaps ###
def all_pages(pdf, whine=lambda *args: None):
return page_descendants(pdf.catalog['Pages'], whine)
@@ -958,6 +918,47 @@ def cmaps_for_pages(pages, whine=lambda *args: None):
else:
whine('not yet handling non-Unicode font', font)
+def buildtables(tokens):
+ "Given a sequence of tokens from a CMap, pull out the tables they represent."
+ csranges = []
+ bfchars = []
+ bfranges = []
+ sizes = {}
+
+ def n_strings_back(n):
+ for j in range(i-n, i):
+ assert tokens[j][0] == 'str'
+ yield tokens[j][1]
+
+ for i, tok in enumerate(tokens):
+ if tok[0] == 'op':
+ op = tok[1]
+ if op in [b'begincodespacerange', b'beginbfchar', b'beginbfrange']:
+ assert tokens[i-1][0] == 'int'
+ sizes[op] = tokens[i-1][1]
+ elif op == b'endcodespacerange':
+ csranges.extend(n_strings_back(2*sizes[b'begincodespacerange']))
+ elif op == b'endbfchar':
+ bfchars.extend(n_strings_back(2*sizes[b'beginbfchar']))
+ elif op == b'endbfrange':
+ bfranges.extend(n_strings_back(3*sizes[b'beginbfrange']))
+
+ return csranges, bfchars, bfranges
+
+
+def csranges_to_grammar(csranges):
+ "Compile the csranges from buildtables into a grammar we can use to tokenize strings."
+ alternatives = []
+ for i in range(0, len(csranges), 2):
+ ranges = zip(csranges[i], csranges[i+1])
+ cs = [Charset(set(c for c in range(startbyte, endbyte+1)))
+ for startbyte, endbyte in ranges]
+ prod = functools.reduce(operator.add, cs)
+ prod.xform = bytes
+ alternatives.append(prod)
+
+ return Any(as_prod(alternatives))
+
if __name__ == '__main__':
import cgitb; cgitb.enable(format='text')
--
GitLab