Work on EOAtable

EditionOpenAccess · Aug 31, 2018 · 91db5d1 · 91db5d1
1 parent e89dc78
commit 91db5d1
Showing 1 changed file with 50 additions and 68 deletions.
diff --git a/tei2imxml.py b/tei2imxml.py
@@ -565,74 +565,52 @@ def transform_body(xml_tree, cited_data, publang):
     ##########
     eoa_tables = xml_tree.xpath("//t:table", namespaces=NS_MAP)
 
-# turn
-            # <table rows="4" cols="4" rend="blank" xml:id="sec3table1">
-            #   <head>This is a table</head>
-            #   <row role="label">
-            #     <cell role="label">Heading 1</cell>
-            #     <cell role="label">Heading <hi rend="bold">2</hi></cell>
-            #     <cell role="label">Heading 3</cell>
-            #     <cell role="label">Heading 4</cell>
-            #   </row>
-            #   <row role="data">
-            #     <cell role="data">Here</cell>
-            #     <cell role="data">you</cell>
-            #     <cell role="data">may</cell>
-            #     <cell role="data">find</cell>
-            #   </row>
-            #   <row role="data">
-            #     <cell role="data">some</cell>
-            #     <cell role="data">data</cell>
-            #     <cell role="data">spread</cell>
-            #     <cell role="data">over</cell>
-            #   </row>
-            #   <row role="data">
-            #     <cell role="data">the</cell>
-            #     <cell role="data">table</cell>
-            #     <cell role="data">in</cell>
-            #     <cell role="data">cells</cell>
-            #   </row>
-            # </table>
-
-# into
-
-# <EOAtable>
-#   <EOAtablelabel>sec3table1</EOAtablelabel>
-#   <EOAtablecaption>This is a table</EOAtablecaption>
-#   <EOAtablecolumns>L2.3cmL2.3cmL2.3cmL2.3cm</EOAtablecolumns>
-#   <table id-text="1" id="uid38" place="sec3table1" rend="display">
-#     <row>
-#       <cell><tableheader>TRUE</tableheader>Heading 1</cell>
-#       <cell>Heading <hi rend="bold">2</hi></cell>
-#       <cell>Heading 3</cell>
-#       <cell>Heading 4</cell>
-#     </row>
-#     <row>
-#       <cell>Here</cell>
-#       <cell>you</cell>
-#       <cell>may</cell>
-#       <cell>find</cell>
-#     </row>
-#     <row>
-#       <cell>some</cell>
-#       <cell>data</cell>
-#       <cell>spread</cell>
-#       <cell>over</cell>
-#     </row>
-#     <row>
-#       <cell>the</cell>
-#       <cell>table</cell>
-#       <cell>in</cell>
-#       <cell>cells</cell>
-#     </row>
-#   </table>
-# </EOAtable>
-
     for table in eoa_tables:
-        table.tag = "EOAtable"
-        table_id = (table.get("{http://www.w3.org/XML/1998/namespace}id"))
+        tablechildren = table.findall("t:row", namespaces=NS_MAP)
+        table_caption = table.find("t:head", namespaces=NS_MAP)
+        number_of_cells = len(table.findall("t:row[1]/t:cell", namespaces=NS_MAP))
+
+        if table_caption is not None:
+            # print(etree.tostring(table))
+            table_id = table.attrib["{http://www.w3.org/XML/1998/namespace}id"]
+            # table_id = table.get("{httbp://www.w3.org/XML/1998/namespace}id")
+            table.clear()
+            table_label = etree.Element("EOAtablelabel")
+            table_label.text = table_id
+            table.append(table_label)
+            table_caption.tag = "EOAtablecaption"
+            table.tag = "EOAtable"
+        else:
+            table.tag = "EOAtablenonumber"
+            table.clear()
+        # not sure if this is evaluated later.
+        table_label = etree.SubElement(table, "EOAtablecolumns").text = "L3cm" * number_of_cells
 
-        table_label = etree.SubElement(table, "EOAtablelabel").text = table_id
+        if table_caption is not None:
+            table.insert(1, table_caption)
+        else:
+            pass
+
+        real_table_element = etree.SubElement(table, "table")
+        real_table_element.set("place", table_id)
+        real_table_element.set("rend", "display")
+        # how are those two constructed?
+        real_table_element.set("id-text", "1")
+        real_table_element.set("id", "uid38")
+
+        for row in tablechildren:
+            if row.get("role") == "label":
+                tableheader = etree.Element("tableheader")
+                tableheader.text = "TRUE"
+                # <tableheader>TRUE</tableheader>
+                row.insert(0, tableheader)
+            else:
+                pass
+            del row.attrib["role"]
+            cells = row.findall("t:cell", namespaces=NS_MAP)
+            for cell in cells:
+                del cell.attrib["role"]
+            real_table_element.append(row)
 
     ##############
     # References #
@@ -747,6 +725,10 @@ def update_ids(xml_tree):
 
     xmlReferences = xml_tree.findall(".//EOAref")
     logging.debug("Found %d references", len(xmlReferences))
+    all_ids = xml_tree.xpath('//*[@xml:id]')
+    for xmlid in all_ids:
+        print(xmlid, "all ids")
+
     for xmlReference in xmlReferences:
         eoa_reference = xmlReference.find("ref")
 
@@ -759,8 +741,8 @@ def update_ids(xml_tree):
         # else:
         corresponding_eoa_id_element = xml_tree.xpath("//*[@xml:id='{}']".format(label_text))
         logging.debug("The corresponding id element is %s", corresponding_eoa_id_element)
-        # if len(corresponding_eoa_id_element) == 0:
-        if corresponding_eoa_id_element is None:
+        # if corresponding_eoa_id_element is None:
+        if len(corresponding_eoa_id_element) == 0:
             print("There seems to be no corresponding xml:id for %s. Exiting." % label_text)
             sys.exit()
         elif len(corresponding_eoa_id_element) > 1: