Word fonts

pdfplumber/utils.py

@@ -1,13 +1,22 @@
 from pdfminer.utils import PDFDocEncoding
 from decimal import Decimal, ROUND_HALF_UP
 import numbers
+from collections import Counter
 from operator import itemgetter
 import itertools
 import six
 
 DEFAULT_X_TOLERANCE = 3
 DEFAULT_Y_TOLERANCE = 3
 
+## Raise an error if the individual characters' font sizes vary by more
+## than this (if we are in strict font height mode )
+DEFAULT_FONT_HEIGHT_TOLERANCE = 0.5
+
+class WordFontError(RuntimeError):
+    def __init__(self,*args,**kwargs):
+        RuntimeError.__init__(self,*args,**kwargs)
+
 def cluster_list(xs, tolerance=0):
     tolerance = decimalize(tolerance)
     if tolerance == 0: return [ [x] for x in sorted(xs) ]
@@ -126,6 +135,51 @@ def objects_to_bbox(objects):
         max(map(itemgetter("bottom"), objects)),
     )
 
+def get_font_from_chars(chars, match_fontname):
+    fontset = set()
+    for char in chars:
+        fontset.add(char['fontname'])
+    number_of_fonts_found = len(fontset)
+    if number_of_fonts_found > 1:
+        if match_fontname:
+            charfonttext = map(itemgetter("fontname"), chars)
+            charlisttext = map(itemgetter("text"), chars)
+            raise WordFontError("Multiple fonts '%s' found in word %s \nPerhaps word tolerance is set too low?" % (charfonttext, charlisttext))
+        return( ", ".join([str(font) for font in fontset]) )
+    if number_of_fonts_found == 0:
+        return ""
+    return fontset.pop()
+
+def get_font_height_from_chars(chars, match_fontsize, font_height_tolerance):
+    charlist = map(itemgetter("height"), chars)
+    max_font_height = max(charlist)
+    min_font_height = min(charlist)
+    font_height_range = max_font_height - min_font_height 
+
+    if match_fontsize and font_height_range > font_height_tolerance:
+        charlist = map(itemgetter("height"), chars)
+        charlisttext = map(itemgetter("text"), chars)
+        raise WordFontError("Font size variation of '%s' exceeds tolerance of %s in word %s with heights %s\nPerhaps word tolerance is set too low?" % (font_height_range, font_height_tolerance, charlisttext, charlist))
+
+    return ( ( max_font_height + min_font_height) / 2 )
+
+def objects_to_bbox_with_font(objects, match_fontname, match_fontsize, font_height_tolerance):
+    fontname = None
+    fontsize = None
+    if match_fontname:
+        fontname = get_font_from_chars(objects, match_fontname)
+    if match_fontsize:
+        fontsize = get_font_height_from_chars(objects, match_fontsize, font_height_tolerance)
+
+    return (
+        min(map(itemgetter("x0"), objects)),
+        min(map(itemgetter("top"), objects)),
+        max(map(itemgetter("x1"), objects)),
+        max(map(itemgetter("bottom"), objects)),
+        fontname,
+        fontsize
+    )
+
 obj_to_bbox = itemgetter("x0", "top", "x1", "bottom")
 
 def bbox_to_rect(bbox):
@@ -139,21 +193,30 @@ def bbox_to_rect(bbox):
 def extract_words(chars,
     x_tolerance=DEFAULT_X_TOLERANCE,
     y_tolerance=DEFAULT_Y_TOLERANCE,
-    keep_blank_chars=False
+    keep_blank_chars=False,
+    match_fontname=True,
+    match_fontsize=True,
+    font_height_tolerance=DEFAULT_FONT_HEIGHT_TOLERANCE
     ):
-
+    
     x_tolerance = decimalize(x_tolerance)
     y_tolerance = decimalize(y_tolerance)
 
     def process_word_chars(chars):
-        x0, top, x1, bottom = objects_to_bbox(chars)
-        return {
+        x0, top, x1, bottom, fontname, fontsize= objects_to_bbox_with_font(chars, match_fontname, match_fontsize, font_height_tolerance)
+        result =  {
             "x0": x0,
             "x1": x1,
             "top": top,
             "bottom": bottom,
-            "text": "".join(map(itemgetter("text"), chars))
+            "text": "".join(map(itemgetter("text"), chars)),
         }
+        if match_fontname:
+            result["fontname"]=fontname
+        if match_fontsize:
+            result["fontsize"]=fontsize
+
+        return result
 
 
     def get_line_words(chars, tolerance=DEFAULT_X_TOLERANCE):

tests/test_fonts.py

@@ -0,0 +1,73 @@
+#!/usr/bin/env python
+import unittest
+import sys
+import os
+import logging
+
+import pdfplumber
+from pdfplumber.utils import extract_words, WordFontError
+
+logging.disable(logging.ERROR)
+HERE = os.path.abspath(os.path.dirname(__file__))
+
+class Test(unittest.TestCase):
+
+    def setUp(self):
+        path = os.path.join(HERE, "pdfs/Acorn_127_201604.pdf")
+        """ 
+        This file has inconsistently named fonts and font sizes.
+        Some words begin with characters roughly 3 (units?) bigger in size.
+        Courtesy of Ryan Ross (Prince Edward Island Guardian)
+        http://www.gov.pe.ca/publicdisclosure/pdSummary.php
+        """
+        test_pdf = pdfplumber.from_path(path)
+        self.pdf_chars = test_pdf.pages[0].chars
+
+        """ As of this writing, the default x_tolerance is 3, y_tolerance is 3 and default_font_height_tolerance is 1 """
+
+        self.default_x_tolerance = 3
+        self.default_y_tolerance = 3
+        self.default_font_height_tolerance = 1
+
+
+    def test_fontname(self):
+        extract_words(self.pdf_chars, 
+            y_tolerance=self.default_y_tolerance, 
+            x_tolerance=self.default_x_tolerance, 
+            font_height_tolerance=self.default_font_height_tolerance, 
+            match_fontsize=False, 
+            match_fontname=False)
+
+        with self.assertRaises(WordFontError):
+            extract_words(self.pdf_chars, 
+                y_tolerance=self.default_y_tolerance, 
+                x_tolerance=self.default_x_tolerance, 
+                font_height_tolerance=self.default_font_height_tolerance, 
+                match_fontsize=False,
+                match_fontname=True)
+
+    def test_font_height_tolerance(self):
+        extract_words(self.pdf_chars, 
+            y_tolerance=self.default_y_tolerance, 
+            x_tolerance=self.default_x_tolerance, 
+            font_height_tolerance=3.5, 
+            match_fontname=False,
+            match_fontsize=True)
+
+        with self.assertRaises(WordFontError): 
+            extract_words(self.pdf_chars, 
+                y_tolerance=self.default_y_tolerance, 
+                x_tolerance=self.default_x_tolerance, 
+                font_height_tolerance=2, 
+                match_fontname=False,
+                match_fontsize=True)
+
+    def test_fontsize(self):
+
+        with self.assertRaises(WordFontError):
+            extract_words(self.pdf_chars, 
+                y_tolerance=self.default_y_tolerance, 
+                x_tolerance=self.default_x_tolerance, 
+                font_height_tolerance=self.default_font_height_tolerance, 
+                match_fontname=False,
+                match_fontsize=True)