updated scripts

scripts/pulltasks.py

@@ -5,6 +5,9 @@ import html5lib
 from xml.etree import ElementTree as ET
 
 
+THUMB_SIZE = 320
+FULL_SIZE = 640
+
 NS_CATEGORY = 14
 
 p = argparse.ArgumentParser(description="Dump wiki files to html")
@@ -51,6 +54,36 @@ def filenameforlink(href):
     path = path+".html"
     return path
 
+def rewriteimagelink(a):
+    href = a.attrib.get("href")
+    path = href
+    if "/" in href:
+        path = path.split("/")[-1]
+    print ("rewriteimagelink", path)
+
+    r = site.api("query", prop="imageinfo", titles=path, iiprop="url", iiurlwidth=str(THUMB_SIZE), formatversion=2)
+    iinfo = r['query']['pages'][0]['imageinfo'][0]
+    thumburl = iinfo['thumburl']
+    #fullsizeurl = iinfo['url']
+    #filepageurl = iinfo['descriptionurl']
+
+    r = site.api("query", prop="imageinfo", titles=path, iiprop="url", iiurlwidth=str(FULL_SIZE), formatversion=2)
+    iinfo = r['query']['pages'][0]['imageinfo'][0]
+    fullsizeurl = iinfo['thumburl']
+    #fullsizeurl = iinfo['url']
+    #filepageurl = iinfo['descriptionurl']
+
+    a.attrib['href'] = fullsizeurl
+    img = a.find("img")
+    img.attrib['src'] = thumburl
+    if "width" in img.attrib:
+        del img.attrib["width"]
+    if "height" in img.attrib:
+        del img.attrib["height"]
+    if "srcset" in img.attrib:
+        del img.attrib["srcset"]
+    print ("rewriteimagelink", thumburl, fullsizeurl)
+
 def rewritelinks(html):
     t = html5lib.parseFragment(html, treebuilder = "etree", namespaceHTMLElements = False)
     for a in t.findall(".//*[@href]"):
@@ -60,7 +93,11 @@ def rewritelinks(html):
             # leave external links alone
             continue
         # print ("LINK", href)
-        if href.startswith("/mediadesign/"):
+        if linkclass == "image":
+            # link to presentation version of image
+            # change img.src to a thumbnail
+            rewriteimagelink(a)
+        elif href.startswith("/mediadesign/"):
             new_href = filenameforlink(href)
             # print ("Rewriting link {} to {}".format(href, new_href), file=sys.stderr)
             a.attrib['href'] = new_href