Ignore path constructors that do not begin with m (#749)

* Ignore path constructors that do not begin with  m

Per PDF Reference Section 4.4.1, "path construction operators may be
invoked in any sequence, but the first one invoked must be m or re to
begin a new subpath." Since pdfminer.six already converts all `re`
(rectangle) operators to their equivelent `mlllh` representation, paths
ingested by `.paint_path(...)` that do not begin with the `m` operator
are invalid.

In addition to the advantage of hewing to the PDF Reference, this change
also avoids the `ValueError: not enough values to unpack (expected 2,
got 1)` error raised by the ` pts = [apply_matrix_pt(self.ctm, pt) for
pt in raw_pts]` line in `converter.py` when parsing PDFs that
(erroneously) include `("h",)` paths.

* Update CHANGELOG.md

Co-authored-by: Pieter Marsman <pietermarsman@gmail.com>

CHANGELOG.md

@@ -8,6 +8,12 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/).
 
 ### Fixed
 
+- Ignoring (invalid) path constructors that do not begin with `m` ([#749](https://github.com/pdfminer/pdfminer.six/pull/749))
+
+## [20220506]
+
+### Fixed
+
 - `IndexError` when handling invalid bfrange code map in
   CMap ([#731](https://github.com/pdfminer/pdfminer.six/pull/731))
 - `TypeError` in lzw.py when `self.table` is not set ([#732](https://github.com/pdfminer/pdfminer.six/pull/732))

pdfminer/converter.py

@@ -109,7 +109,16 @@ class PDFLayoutAnalyzer(PDFTextDevice):
         """Paint paths described in section 4.4 of the PDF reference manual"""
         shape = "".join(x[0] for x in path)
 
-        if shape.count("m") > 1:
+        if shape[:1] != "m":
+            # Per PDF Reference Section 4.4.1, "path construction operators may
+            # be invoked in any sequence, but the first one invoked must be m
+            # or re to begin a new subpath." Since pdfminer.six already
+            # converts all `re` (rectangle) operators to their equivelent
+            # `mlllh` representation, paths ingested by `.paint_path(...)` that
+            # do not begin with the `m` operator are invalid.
+            pass
+
+        elif shape.count("m") > 1:
             # recurse if there are multiple m's in this shape
             for m in re.finditer(r"m[^m]+", shape):
                 subpath = path[m.start(0) : m.end(0)]

tests/test_converter.py

@@ -215,6 +215,15 @@ class TestPaintPath:
             (71.41, 434.89),
         ]
 
+    def test_paint_path_without_starting_m(self):
+        gs = PDFGraphicState()
+        analyzer = self._get_analyzer()
+        analyzer.cur_item = LTContainer([0, 100, 0, 100])
+        paths = [[("h",)], [("l", 72.41, 433.89), ("l", 82.41, 433.89), ("h",)]]
+        for path in paths:
+            analyzer.paint_path(gs, False, False, False, path)
+        assert len(analyzer.cur_item._objs) == 0
+
 
 class TestBinaryDetector:
     def test_stringio(self):