fix: Change ocr_line <span> to include all ocr_word (#169)

Fixes the xml for ocr_line. The span of ocr_line should enclose all spans of ocr_word
googleapis · Oct 3, 2023 · bc44dab · bc44dab
1 parent 3c3f09d
commit bc44dab
Show file tree

Hide file tree

Showing 2 changed files with 77 additions and 76 deletions.
diff --git a/google/cloud/documentai_toolbox/templates/hocr_document_template.xml.j2 b/google/cloud/documentai_toolbox/templates/hocr_document_template.xml.j2
@@ -18,9 +18,10 @@
             {% set paridx = loop.index0 -%}
             <span class='ocr_par' id='par_{{ page_number }}_{{ bidx }}_{{ paridx }}' title='{{ paragraph.hocr_bounding_box -}}'>{% for line in paragraph.lines -%}
                 {% set lidx = loop.index0 -%}
-                <span class='ocr_line' id='line_{{ page_number }}_{{ bidx }}_{{ paridx }}_{{ lidx }}' title='{{ line.hocr_bounding_box }}'>{{ line.text }}</span>{% for token in line.tokens -%}
+                <span class='ocr_line' id='line_{{ page_number }}_{{ bidx }}_{{ paridx }}_{{ lidx }}' title='{{ line.hocr_bounding_box }}'>{{ line.text }}{% for token in line.tokens -%}
                     {% set tidx = loop.index0 -%}
-                    <span class='ocrx_word' id='word_{{ page_number }}_{{ bidx }}_{{ paridx }}_{{ lidx }}_{{ tidx }}' title='{{ token.hocr_bounding_box }}'>{{ token.text }}</span>{% endfor -%}{% endfor -%}
+                    <span class='ocrx_word' id='word_{{ page_number }}_{{ bidx }}_{{ paridx }}_{{ lidx }}_{{ tidx }}' title='{{ token.hocr_bounding_box }}'>{{ token.text }}</span>{% endfor -%}
+                </span>{% endfor -%}
             </span>{% endfor -%}
         </span>{% endfor -%}
     </div>