From 20a21c37f1ab54f040da338e3d89200d8d30c876 Mon Sep 17 00:00:00 2001
From: kthoden <kthoden@gwdg.de>
Date: Wed, 31 Jul 2019 14:58:26 +0200
Subject: [PATCH] Hyperimage figure handling, mostly collages

---
 imxml2django.py | 50 ++++++++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 47 insertions(+), 3 deletions(-)

diff --git a/imxml2django.py b/imxml2django.py
index 49ec48c..0b82359 100755
--- a/imxml2django.py
+++ b/imxml2django.py
@@ -1,6 +1,6 @@
 #!/usr/bin/env python3
 # -*- coding: utf-8; mode: python -*-
-# Time-stamp: <2019-07-30 16:05:12 (kthoden)>
+# Time-stamp: <2019-07-31 14:46:18 (kthoden)>
 
 """
 Create an XML file that can be inserted into the Django database
@@ -387,7 +387,7 @@ def djangoParseObject(xmlElement, indent=False, listtype=None, listnumber=0, uid
             figure_type = xmlElement.get("type")
             # Copy Image
             if figure_type == "hionly" or figure_type == "hionlycollage":
-                logging.debug(f"Found hyperimage figure ({figure_type}), continuing")
+                logging.debug(f"Found hyperimage figure ({figure_type}), no need to copy them.")
                 pass
             else:
                 strImageFileString = xmlElement.find(".//file").text
@@ -415,17 +415,29 @@ def djangoParseObject(xmlElement, indent=False, listtype=None, listnumber=0, uid
                     logging.debug("The filename is %s" % xmlEOAfigure.get("file"))
                 else:
                     xmlEOAfigure.set("file", strImageFileDir + strImageFileName)
+
+            if figure_type == "hionly" or figure_type == "hionlycollage":
+                logging.debug(f"Found hyperimage figure ({figure_type}), no need for caption and size information.")
+                pass
+            else:
                 xmlEOAfigure.set("width", xmlElement.find(".//width").text + "px;")
                 xmlEOAfigure.set("order", str(intObjectNumber))
                 xmlEOAfigure.append(xmlElement.find(".//caption"))
                 xmlResult.append(xmlEOAfigure)
-
                 intObjectNumber += 1
                 # Insert visual Number and uid
                 strFigureNumber = dictFigures[xmlElement.find(".//anchor").get("id")]
                 xmlEOAfigure.set("number", strFigureNumber)
                 strFigureUID = xmlElement.find(".//anchor").get("id")
                 xmlEOAfigure.set("id", strFigureUID)
+
+            hi_figure_types = ["hitrue", "hionly", "hionlycollage"]
+
+            if figure_type in hi_figure_types:
+                xmlEOAfigure.set("hielement", xmlElement.get("hielement"))
+            else:
+                pass
+
         elif xmlElement.findall(".//EOAtable"):
             xmlResult = etree.Element("EOAtable")
             xmlRawTable = xmlElement.find(".//table")
@@ -1512,6 +1524,27 @@ def bring_footnote_down_django(footnote, fragment, footnote_number, object_numbe
 
 # Substitute References with their targets (wit links)
 for xmlEOAchapter in xmlEOAchapters:
+    # for hyperimage collages
+    originalcontents = xmlEOAchapter.findall(".//originalcontents")
+    if originalcontents is not None:
+        for originalcontent in originalcontents:
+            previous_element = originalcontent.getprevious()
+            if originalcontent.getparent().tag == "EOAref":
+                pass
+            elif previous_element.tag != "EOAref":
+                logging.error("Found a stray originalcontents element.")
+            else:
+                oc_tail = originalcontent.tail
+                originalcontent.tail = ""
+                previous_element.append(originalcontent)
+                if previous_element.tail is not None:
+                    logging.warning("Appending the old tail of EOAref")
+                    previous_element.tail += oc_tail
+                else:
+                    previous_element.tail = oc_tail
+    else:
+        logging.debug("No originalcontents elements found.")
+
     xmlReferences = xmlEOAchapter.findall(".//EOAref")
     for xmlReference in xmlReferences:
         strResult = "!!! Cross Reference !!!"
@@ -1633,6 +1666,10 @@ def bring_footnote_down_django(footnote, fragment, footnote_number, object_numbe
         tmpTail = xmlReference.tail or ""
 
         originalcontents = xmlReference.find("originalcontents")
+        if xmlReference.get("type") == "collage":
+            ref_is_collage = True
+        else:
+            ref_is_collage = False
         xmlReference.clear()
         if originalcontents is not None:
             logging.info("Found originalcontents")
@@ -1658,6 +1695,10 @@ def bring_footnote_down_django(footnote, fragment, footnote_number, object_numbe
             logging.warning("strObjectOrder is missing!")
 
         xmlReference.set("href", href_string)
+        if ref_is_collage:
+            xmlReference.set("type", "collage")
+        else:
+            pass
 
 logging.info("----------------------------------------------")
 logging.info("Processing Page References")
@@ -1857,6 +1898,9 @@ def bring_footnote_down_django(footnote, fragment, footnote_number, object_numbe
 ############################################################################
 # TODO: Die unnötigen Attribute wie id löschen
 # TODO: Die unnötigen Tags wie EOAlabel löschen
+collagelinks = xmlDjangoTree.xpath(".//a[@type='collage']/originalcontents/a")
+for link in collagelinks:
+    link.tag = "temp"
 etree.strip_tags(xmlDjangoTree, "temp", "citetext", "EOAprintbibliography", "originalcontents")
 etree.strip_elements(xmlDjangoTree, "citekey", with_tail=False)
 etree.strip_attributes(xmlDjangoTree, "id-text", "id", "noindent", "type", "label", "spacebefore")#, "rend")