consistent test results

git-svn-id: https://pdfminerr.googlecode.com/svn/trunk/pdfminer@191 1aa58f4a-7d42-0410-adbc-911cccaed67c
phoenixlqh · Mar 22, 2010 · 40b36a7 · 40b36a7
1 parent a6523d1
commit 40b36a7
Show file tree

Hide file tree

Showing 20 changed files with 13,059 additions and 13,043 deletions.
diff --git a/Makefile b/Makefile
@@ -50,6 +50,6 @@ $(CMAPDST)/TO_UNICODE_Adobe_Korea1.py:
 	$(CONV_CMAP) $(CMAPDST) Adobe-Korea1 $(CMAPSRC)/cid2code_Adobe_Korea1.txt cp949 euc-kr
 
 test: cmap
-	cd samples && $(MAKE) test
+	cd samples && $(MAKE) test CMP=cmp
 test_clean:
 	-cd samples && $(MAKE) clean
diff --git a/pdfminer/layout.py b/pdfminer/layout.py
@@ -27,6 +27,16 @@ def uniq(objs):
         yield obj
     return
 
+def csort(objs, key):
+    idxs = dict( (obj,i) for (i,obj) in enumerate(objs) )
+    return sorted(objs, key=lambda obj:(key(obj), idxs[obj]))
+
+def is_uniq(objs):
+    for (i,obj1) in enumerate(objs):
+        for obj2 in objs[i+1:]:
+            if obj1 == obj2: return False
+    return True
+
 
 ##  LAParams
 ##
@@ -324,7 +334,7 @@ def __init__(self, objs, word_margin):
         LayoutContainer.fixate(self)
         objs = []
         x1 = INF
-        for obj in sorted(self.objs, key=lambda obj: obj.x0):
+        for obj in csort(self.objs, key=lambda obj: obj.x0):
             if isinstance(obj, LTChar) and word_margin:
                 margin = word_margin * obj.width
                 if x1 < obj.x0-margin:
@@ -345,7 +355,7 @@ def __init__(self, objs, word_margin):
         LayoutContainer.fixate(self)
         objs = []
         y0 = -INF
-        for obj in sorted(self.objs, key=lambda obj: -obj.y1):
+        for obj in csort(self.objs, key=lambda obj: -obj.y1):
             if isinstance(obj, LTChar) and word_margin:
                 margin = word_margin * obj.height
                 if obj.y1+margin < y0:
@@ -382,14 +392,14 @@ class LTTextBoxHorizontal(LTTextBox):
 
     def fixate(self):
         LTTextBox.fixate(self)
-        self.objs = sorted(self.objs, key=lambda obj: -obj.y1)
+        self.objs = csort(self.objs, key=lambda obj: -obj.y1)
         return
 
 class LTTextBoxVertical(LTTextBox):
 
     def fixate(self):
         LTTextBox.fixate(self)
-        self.objs = sorted(self.objs, key=lambda obj: -obj.x1)
+        self.objs = csort(self.objs, key=lambda obj: -obj.x1)
         return
 
 
@@ -408,15 +418,15 @@ class LTTextGroupHorizontal(LTTextGroup):
     def __init__(self, objs):
         LTTextGroup.__init__(self, objs)
         # reorder the objects from top-left to bottom-right.
-        self.objs = sorted(self.objs, key=lambda obj: obj.x0-obj.y1)
+        self.objs = csort(self.objs, key=lambda obj: obj.x0-obj.y1)
         return
 
 class LTTextGroupVertical(LTTextGroup):
 
     def __init__(self, objs):
         LTTextGroup.__init__(self, objs)
         # reorder the objects from top-right to bottom-left.
-        self.objs = sorted(self.objs, key=lambda obj: -obj.x1-obj.y1)
+        self.objs = csort(self.objs, key=lambda obj: -obj.x1-obj.y1)
         return
 
 
@@ -432,6 +442,7 @@ class Plane(object):
     def __init__(self, objs):
         self.xobjs = []
         self.yobjs = []
+        self.idxs = dict( (obj,i) for (i,obj) in enumerate(objs) )
         for obj in objs:
             self.place(obj)
         self.xobjs.sort()
@@ -456,7 +467,7 @@ def find(self, (x0,y0,x1,y1)):
         i1 = bsearch(self.yobjs, y1)[1]
         yobjs = set( obj for (_,obj) in self.yobjs[i0:i1] )
         xobjs.intersection_update(yobjs)
-        return list(xobjs)
+        return sorted(xobjs, key=lambda obj: self.idxs[obj])
 
 
 ##  group_lines
@@ -474,21 +485,25 @@ def group_lines(groupfunc, objs, *args):
         group = groupfunc(list(uniq(members)))
         for obj in members:
             groups[obj] = group
-    groups = set(groups.values())
-    for group in groups:
+    done = set()
+    r = []
+    for obj in objs:
+        group = groups[obj]
+        if group in done: continue
+        done.add(group)
         group.fixate()
-    return list(groups)
+        r.append(group)
+    return r
 
 
 ##  group_boxes
 ##
 def group_boxes(groupfunc, objs, distfunc):
     assert objs
-    objs = objs[:]
     while 2 <= len(objs):
         mindist = INF
         minpair = None
-        objs.sort(key=lambda obj: (obj.width*obj.height, obj.y0))
+        objs = csort(objs, key=lambda obj: obj.width*obj.height)
         for i in xrange(len(objs)):
             for j in xrange(i+1, len(objs)):
                 d = distfunc(objs[i], objs[j])

diff --git a/samples/Makefile b/samples/Makefile
@@ -1,7 +1,8 @@
 # GNUMakefile for test
 
 RM=rm -f
-CMP=cmp
+#CMP=cmp
+CMP=:
 PYTHON=python
 PDF2TXT=PYTHONPATH=.. $(PYTHON) ../tools/pdf2txt.py -p1
 

diff --git a/samples/dmca.html.ref b/samples/dmca.html.ref
@@ -1945,7 +1945,7 @@
 <span style="position:absolute; border: red 1px solid; left:143px; top:379px; width:8px; height:214px;"></span>
 <span style="position:absolute; border: red 1px solid; left:143px; top:379px; width:8px; height:133px;"></span>
 <span style="position:absolute; border: red 1px solid; left:144px; top:379px; width:8px; height:93px;"></span>
-<span style="position:absolute; border: red 1px solid; left:144px; top:419px; width:8px; height:52px;"></span>
+<span style="position:absolute; border: red 1px solid; left:144px; top:379px; width:8px; height:52px;"></span>
 <span style="position:absolute; border: red 1px solid; left:108px; top:726px; width:396px; height:37px;"></span>
 <span style="position:absolute; border: red 1px solid; left:108px; top:750px; width:396px; height:13px;"></span>
 <span style="position:absolute; border: red 1px solid; left:108px; top:750px; width:229px; height:13px;"></span>

diff --git a/samples/dmca.xml.ref b/samples/dmca.xml.ref
@@ -2197,11 +2197,11 @@
 <textgroup bbox="144.000,248.400,152.004,462.480">
 <textgroup bbox="144.000,329.040,152.004,462.480">
 <textgroup bbox="144.000,369.360,152.004,462.480">
+<textgroup bbox="144.000,409.680,152.004,462.480">
 <textbox id="5" bbox="144.000,450.000,152.004,462.480" />
-<textgroup bbox="144.000,369.360,152.004,422.160">
 <textbox id="6" bbox="144.000,409.680,152.004,422.160" />
-<textbox id="7" bbox="144.000,369.360,152.004,381.840" />
 </textgroup>
+<textbox id="7" bbox="144.000,369.360,152.004,381.840" />
 </textgroup>
 <textbox id="8" bbox="144.000,329.040,152.004,341.520" />
 </textgroup>

diff --git a/samples/f1040nr.html.ref b/samples/f1040nr.html.ref
@@ -568,6 +568,14 @@
 <span style="position:absolute; left:425px; top:184px; font-size:6px;">'</span>
 <span style="position:absolute; border: blue 1px solid; left:56px; top:193px; width:502px; height:40px;"></span>
 <span style="position:absolute; left:425px; top:193px; font-size:5px;"> </span>
+<span style="position:absolute; left:56px; top:197px; font-size:9px;">C</span>
+<span style="position:absolute; left:62px; top:197px; font-size:9px;">o</span>
+<span style="position:absolute; left:66px; top:197px; font-size:9px;">u</span>
+<span style="position:absolute; left:71px; top:197px; font-size:9px;">n</span>
+<span style="position:absolute; left:75px; top:197px; font-size:9px;">t</span>
+<span style="position:absolute; left:78px; top:197px; font-size:9px;">r</span>
+<span style="position:absolute; left:80px; top:197px; font-size:9px;">y</span>
+<span style="position:absolute; left:88px; top:196px; font-size:7px;">'</span>
 <span style="position:absolute; left:251px; top:197px; font-size:9px;">O</span>
 <span style="position:absolute; left:256px; top:197px; font-size:9px;">f</span>
 <span style="position:absolute; left:257px; top:197px; font-size:9px;"> </span>
@@ -636,14 +644,6 @@
 <span style="position:absolute; left:430px; top:197px; font-size:9px;">r</span>
 <span style="position:absolute; left:432px; top:197px; font-size:9px;">?</span>
 <span style="position:absolute; left:439px; top:196px; font-size:7px;">'</span>
-<span style="position:absolute; left:56px; top:197px; font-size:9px;">C</span>
-<span style="position:absolute; left:62px; top:197px; font-size:9px;">o</span>
-<span style="position:absolute; left:66px; top:197px; font-size:9px;">u</span>
-<span style="position:absolute; left:71px; top:197px; font-size:9px;">n</span>
-<span style="position:absolute; left:75px; top:197px; font-size:9px;">t</span>
-<span style="position:absolute; left:78px; top:197px; font-size:9px;">r</span>
-<span style="position:absolute; left:80px; top:197px; font-size:9px;">y</span>
-<span style="position:absolute; left:88px; top:196px; font-size:7px;">'</span>
 <span style="position:absolute; left:88px; top:207px; font-size:8px;"> </span>
 <span style="position:absolute; left:439px; top:207px; font-size:8px;"> </span>
 <span style="position:absolute; left:56px; top:208px; font-size:9px;">G</span>
@@ -3783,16 +3783,16 @@
 <span style="position:absolute; left:358px; top:473px; font-size:10px;">9</span>
 <span style="position:absolute; left:358px; top:474px; font-size:10px;"> </span>
 <span style="position:absolute; left:363px; top:473px; font-size:10px;">b</span>
-<span style="position:absolute; border: blue 1px solid; left:318px; top:727px; width:2px; height:10px;"></span>
-<span style="position:absolute; left:318px; top:727px; font-size:10px;"> </span>
-<span style="position:absolute; border: blue 1px solid; left:308px; top:691px; width:2px; height:10px;"></span>
-<span style="position:absolute; left:308px; top:691px; font-size:10px;"> </span>
 <span style="position:absolute; border: blue 1px solid; left:435px; top:607px; width:2px; height:10px;"></span>
 <span style="position:absolute; left:435px; top:607px; font-size:10px;"> </span>
-<span style="position:absolute; border: blue 1px solid; left:256px; top:571px; width:2px; height:10px;"></span>
-<span style="position:absolute; left:256px; top:571px; font-size:10px;"> </span>
+<span style="position:absolute; border: blue 1px solid; left:308px; top:691px; width:2px; height:10px;"></span>
+<span style="position:absolute; left:308px; top:691px; font-size:10px;"> </span>
+<span style="position:absolute; border: blue 1px solid; left:318px; top:727px; width:2px; height:10px;"></span>
+<span style="position:absolute; left:318px; top:727px; font-size:10px;"> </span>
 <span style="position:absolute; border: blue 1px solid; left:281px; top:486px; width:2px; height:10px;"></span>
 <span style="position:absolute; left:281px; top:486px; font-size:10px;"> </span>
+<span style="position:absolute; border: blue 1px solid; left:256px; top:571px; width:2px; height:10px;"></span>
+<span style="position:absolute; left:256px; top:571px; font-size:10px;"> </span>
 <span style="position:absolute; border: blue 1px solid; left:63px; top:437px; width:5px; height:11px;"></span>
 <span style="position:absolute; left:63px; top:437px; font-size:10px;">d</span>
 <span style="position:absolute; left:63px; top:438px; font-size:10px;"> </span>

diff --git a/samples/f1040nr.txt.ref b/samples/f1040nr.txt.ref
@@ -43,8 +43,8 @@ Type  of  entry  visa  (see  page  8)
 '
 
 
-Of what  country were  you  a citizen or  national  during  the  tax  year? '
 Country '
+Of what  country were  you  a citizen or  national  during  the  tax  year? '
 
 
 Give  address outside  the United  States to which  you want  any

diff --git a/samples/f1040nr.xml.ref b/samples/f1040nr.xml.ref
@@ -801,6 +801,19 @@
 <text>
 </text>
 </textline>
+<textline bbox="56.488,635.344,95.968,645.365">
+<text font="HelveticaNeue-Roman" vertical="False" bbox="56.488,635.344,62.264,644.672" size="9.328">C</text>
+<text font="HelveticaNeue-Roman" vertical="False" bbox="62.264,635.344,66.856,644.672" size="9.328">o</text>
+<text font="HelveticaNeue-Roman" vertical="False" bbox="66.856,635.344,71.304,644.672" size="9.328">u</text>
+<text font="HelveticaNeue-Roman" vertical="False" bbox="71.304,635.344,75.752,644.672" size="9.328">n</text>
+<text font="HelveticaNeue-Roman" vertical="False" bbox="75.752,635.344,78.272,644.672" size="9.328">t</text>
+<text font="HelveticaNeue-Roman" vertical="False" bbox="78.272,635.344,80.936,644.672" size="9.328">r</text>
+<text font="HelveticaNeue-Roman" vertical="False" bbox="80.936,635.344,84.936,644.672" size="9.328">y</text>
+<text> </text>
+<text font="Universal-NewswithCommPi" vertical="False" bbox="88.968,638.428,95.968,645.365" size="7.000">'</text>
+<text>
+</text>
+</textline>
 <textline bbox="251.598,635.344,446.738,645.365">
 <text font="HelveticaNeue-Roman" vertical="False" bbox="251.598,635.344,256.158,644.672" size="9.328">O</text>
 <text font="HelveticaNeue-Roman" vertical="False" bbox="256.158,635.344,257.934,644.672" size="9.328">f</text>
@@ -882,19 +895,6 @@
 <text>
 </text>
 </textline>
-<textline bbox="56.488,635.344,95.968,645.365">
-<text font="HelveticaNeue-Roman" vertical="False" bbox="56.488,635.344,62.264,644.672" size="9.328">C</text>
-<text font="HelveticaNeue-Roman" vertical="False" bbox="62.264,635.344,66.856,644.672" size="9.328">o</text>
-<text font="HelveticaNeue-Roman" vertical="False" bbox="66.856,635.344,71.304,644.672" size="9.328">u</text>
-<text font="HelveticaNeue-Roman" vertical="False" bbox="71.304,635.344,75.752,644.672" size="9.328">n</text>
-<text font="HelveticaNeue-Roman" vertical="False" bbox="75.752,635.344,78.272,644.672" size="9.328">t</text>
-<text font="HelveticaNeue-Roman" vertical="False" bbox="78.272,635.344,80.936,644.672" size="9.328">r</text>
-<text font="HelveticaNeue-Roman" vertical="False" bbox="80.936,635.344,84.936,644.672" size="9.328">y</text>
-<text> </text>
-<text font="Universal-NewswithCommPi" vertical="False" bbox="88.968,638.428,95.968,645.365" size="7.000">'</text>
-<text>
-</text>
-</textline>
 <textline bbox="88.968,626.542,90.914,634.704">
 <text font="HelveticaNeue-Roman" vertical="False" bbox="88.968,626.542,90.914,634.704" size="8.162"> </text>
 <text>
@@ -5326,9 +5326,9 @@
 </text>
 </textline>
 </textbox>
-<textbox id="29" bbox="318.048,103.606,320.550,114.100">
-<textline bbox="318.048,103.606,320.550,114.100">
-<text font="HelveticaNeue-Roman" vertical="False" bbox="318.048,103.606,320.550,114.100" size="10.494"> </text>
+<textbox id="29" bbox="435.798,224.046,438.300,234.540">
+<textline bbox="435.798,224.046,438.300,234.540">
+<text font="HelveticaNeue-Roman" vertical="False" bbox="435.798,224.046,438.300,234.540" size="10.494"> </text>
 <text>
 </text>
 </textline>
@@ -5340,23 +5340,23 @@
 </text>
 </textline>
 </textbox>
-<textbox id="31" bbox="435.798,224.046,438.300,234.540">
-<textline bbox="435.798,224.046,438.300,234.540">
-<text font="HelveticaNeue-Roman" vertical="False" bbox="435.798,224.046,438.300,234.540" size="10.494"> </text>
+<textbox id="31" bbox="318.048,103.606,320.550,114.100">
+<textline bbox="318.048,103.606,320.550,114.100">
+<text font="HelveticaNeue-Roman" vertical="False" bbox="318.048,103.606,320.550,114.100" size="10.494"> </text>
 <text>
 </text>
 </textline>
 </textbox>
-<textbox id="32" bbox="256.488,260.196,258.990,270.690">
-<textline bbox="256.488,260.196,258.990,270.690">
-<text font="HelveticaNeue-Roman" vertical="False" bbox="256.488,260.196,258.990,270.690" size="10.494"> </text>
+<textbox id="32" bbox="281.508,344.536,284.010,355.030">
+<textline bbox="281.508,344.536,284.010,355.030">
+<text font="HelveticaNeue-Roman" vertical="False" bbox="281.508,344.536,284.010,355.030" size="10.494"> </text>
 <text>
 </text>
 </textline>
 </textbox>
-<textbox id="33" bbox="281.508,344.536,284.010,355.030">
-<textline bbox="281.508,344.536,284.010,355.030">
-<text font="HelveticaNeue-Roman" vertical="False" bbox="281.508,344.536,284.010,355.030" size="10.494"> </text>
+<textbox id="33" bbox="256.488,260.196,258.990,270.690">
+<textline bbox="256.488,260.196,258.990,270.690">
+<text font="HelveticaNeue-Roman" vertical="False" bbox="256.488,260.196,258.990,270.690" size="10.494"> </text>
 <text>
 </text>
 </textline>
@@ -5981,15 +5981,15 @@
 </textgroup>
 <textbox id="28" bbox="358.438,356.790,368.941,368.427" />
 </textgroup>
-<textbox id="29" bbox="318.048,103.606,320.550,114.100" />
+<textbox id="29" bbox="435.798,224.046,438.300,234.540" />
 </textgroup>
 <textbox id="30" bbox="308.058,139.676,310.560,150.170" />
 </textgroup>
-<textbox id="31" bbox="435.798,224.046,438.300,234.540" />
+<textbox id="31" bbox="318.048,103.606,320.550,114.100" />
 </textgroup>
-<textbox id="32" bbox="256.488,260.196,258.990,270.690" />
+<textbox id="32" bbox="281.508,344.536,284.010,355.030" />
 </textgroup>
-<textbox id="33" bbox="281.508,344.536,284.010,355.030" />
+<textbox id="33" bbox="256.488,260.196,258.990,270.690" />
 </textgroup>
 <textbox id="34" bbox="63.688,392.640,69.187,404.277" />
 </textgroup>