fixes #6 and fixes #7

ceurspt/__init__.py

@@ -1 +1 @@
-__version__="0.0.2"
+__version__="0.0.3"

ceurspt/ceurws.py

@@ -5,20 +5,61 @@
 '''
 import ceurspt.ceurws_base
 import os
+import re
+from bs4 import BeautifulSoup
+
+class Paper(ceurspt.ceurws_base.Paper):
+    """
+    a CEUR-WS Paper with it's behavior
+    """
+
+    def getPdf(self):
+        """
+        get the PDF file for this paper
+        """
+        volume=self.volume
+        pdf=f"{volume.vm.base_path}/Vol-{volume.number}/paper{self.paper_number}.pdf"
+        return pdf
 
 class Volume(ceurspt.ceurws_base.Volume):
     """
     a CEUR-WS Volume with it's behavior
     """
 
-    def getHtml(self)->str:
+    def getHtml(self,fixLinks:bool=True)->str:
         """
         get my HTML content
+        
+        Args:
+            fixLinks(bool): if True fix the links
         """
         index_path=f"{self.vol_dir}/index.html"
         with open(index_path, 'r') as index_html:
             content = index_html.read()
+            if fixLinks:
+                soup = BeautifulSoup(content, 'html.parser')
+                for a in soup.findAll(['link','a']):
+                    ohref=a['href'] 
+                    # .replace("google", "mysite")
+                    href=ohref.replace("http://ceur-ws.org/","/")
+                    href=href.replace("../ceur-ws.css","/ceur-ws.css")
+                    href=re.sub(r'paper([0-9]+).pdf', fr"/Vol-{self.number}/paper\g<1>.pdf", href)
+                    pass
+                    a['href']=href
+                content=soup.prettify( formatter="html" )
             return content
+
+    def getPaper(self,paper_number:int):
+        """
+        get the paper with the given number
+        
+        Args:
+            paper_number(int): the number of the paper
+        """
+        paper=Paper()
+        paper.paper_number=paper_number
+        paper.volume=self
+        return paper
 
 class VolumeManager():
     """
@@ -46,6 +87,7 @@ def getVolume(self,number:int)->Volume:
         vol_dir=f"{self.base_path}/Vol-{number}"
         if os.path.isdir(vol_dir):
             vol=Volume(number=number)
+            vol.number=int(vol.number)
             vol.vm=self
             vol.vol_dir=vol_dir
             return vol

ceurspt/webserver.py

@@ -4,7 +4,8 @@
 @author: wf
 """
 from fastapi import FastAPI
-from fastapi.responses import HTMLResponse, RedirectResponse
+from fastapi.responses import FileResponse, HTMLResponse, RedirectResponse
+from fastapi.staticfiles import StaticFiles
 from ceurspt.ceurws import VolumeManager
 import dataclasses
 
@@ -13,14 +14,32 @@ class WebServer:
     the webserver
     """
 
-    def __init__(self,vm:VolumeManager):
+    def __init__(self,vm:VolumeManager,static_directory:str="static"):
         """
         constructor
+        
+        Args:
+            vm(VolumeManager): the volume manager to use
+            static_directory(str): the directory for static html files to use
         """
         self.app = FastAPI()
+        #https://fastapi.tiangolo.com/tutorial/static-files/
+        self.app.mount("/static", StaticFiles(directory=static_directory), name="static")
         self.vm=vm
 
 
+        @self.app.get("/Vol-{number:int}/paper{paper_number:int}.pdf")
+        async def paperPdf(number:int,paper_number:int):
+            """
+            get the PDF for the given paper
+            """
+            vol=self.vm.getVolume(number)
+            paper=vol.getPaper(paper_number)
+            pdf=paper.getPdf()
+            return FileResponse(pdf)
+
+
+
         @self.app.get("/Vol-{number:int}.json")
         async def volumeJson(number: int):
             """
@@ -40,7 +59,7 @@ async def volumeHtml(number:int):
             """
             vol=self.vm.getVolume(number)
             if vol:
-                content=vol.getHtml()
+                content=vol.getHtml(fixLinks=True)
                 return HTMLResponse(content=content, status_code=200)
             else:
                 return HTMLResponse(content=f"unknown volume number {number}",status_code=404)

pyproject.toml

@@ -21,7 +21,9 @@ dependencies = [
   # https://pypi.org/project/uvicorn/
   "uvicorn",
   # https://pypi.org/project/linkml-runtime/
-  "linkml-runtime"
+  "linkml-runtime",
+  # https://pypi.org/project/beautifulsoup4/
+  "beautifulsoup4"
 ]
 
 requires-python = ">=3.8"

static/ceur-ws.css

@@ -0,0 +1,135 @@
+/* ceur-ws.css defines how links are displayed and        */
+/* possibly some other details of page rendering within   */
+/* CEUR-WS.org.                                           */
+/* Copyright 2002-2015 by Manfred Jeusfeld                */
+/* Use permitted within CEUR-WS.org.                      */
+/* Modification permitted for members of the CEUR-WS TEAM.*/
+/*                                                        */
+/* THIS STYLE FILE IS FOR EXCLUSIVE USE WITHIN            */
+/* CEUR-WS.ORG! DO NOT COPY IT AS A WHOLE OR AS PARTS.    */
+/* IN PARTICULAR, THE TAGS CEUR* DEFINED BELOW MAY NOT BE */
+/* REUSED OR OVERRIDDEN.                                  */
+/* All references to ceur-ws.css in HTML files must be    */
+/* relative to ceur-ws.org, e.g. "../ceur-ws.ccs".        */
+/* Absolute references like "ceur-ws.org/ceur-ws.css" are */
+/* not permitted.                                         */
+/*                                                        */
+/* Created: 5-Nov-2002/M.Jeusfeld                         */
+/* Modified: 5-Sep-2005/M.Jeusfeld                        */
+/* Modified: 22-Nov-2006/M.Jeusfeld  (hover darker)       */
+/* Modified: 17-Dec-2008/M.Jeusfeld  (specific font fam.) */
+/* Modified: 15-Jan-2010/M.Jeusfeld  (CEUR class tags)    */
+/* Modified: 14-Apr-2010/M.Jeusfeld  (adapt link colors)  */
+/* Modified:  7-Jul-2010/M.Jeusfeld  (AUX tags)           */
+/* Modified: 13-Oct-2011/M.Jeusfeld  (ceur-ws.org)        */
+/* Modified: 20-Jun-2012/M.Jeusfeld  (color of VOLNR,URN) */
+/* Modified: 02-Jul-2012/M.Jeusfeld  (square bullet LI)   */
+/* Modified: 08-Jul-2013/C.Lange     (.unobtrusive and .copyright, to get rid of <font>) */
+/* Modified: 16-Oct-2013/M.Jeusfeld  (use grey for blue)  */
+/* Modified: 11-Dec-2013/M.Jeusfeld  (CEURWSLOGO div)     */
+/* Modified: 24-Mar-2015/M.Jeusfeld  (paper directories)  */
+/* Modified: 16-Jul-2015/C.Lange     (more accessible link colors) */
+/* Modified: 08-Feb-2016/M.Jeusfeld  (distinguish AUX from CEURTITLES) */
+/* Modified: 05-Oct-2019/M.Jeusfeld  (CEURCCBY tag) */
+/* Modified: 30-Oct-2019/M.Jeusfeld  (write before Volnr disabled) */
+/* Modified: 03-Dec-2019/M.Jeusfeld  (print bg images) */
+/* Modified: 07-Nov-2022/M.Jeusfeld  (use https) */
+
+
+body {
+font-family: arial, sans-serif;
+font-size: medium;
+background-image: url(/static/ceurws-bg.png);
+margin-left: +32px;
+margin-right: +8px;
+}
+
+/* 2015-04-21/MJ: for mobile-friendly layout */
+@-ms-viewport{
+  width: device-width;
+  initial-scale: 1.0;
+}
+
+@media print {
+    body > header address {
+        display:none;
+    }
+    body > header #document-license {
+        width:19em;
+        clear:right;
+        float:right;
+        margin-top:1em !important;
+        margin-right:0;
+    }
+    #content {
+        page-break-before:always;
+    }
+}
+
+A:link {color: #404080; text-decoration: none;}  /* display of links */
+A:visited {color: #404080; text-decoration: none;} /* visited links */
+A:hover {color: #000080; text-decoration: none;} /* links under the cursor */
+:target {background-color:#FFFFFF;} /* background for current fragment */
+
+H1 {color:#000000; font-family:arial, cursive, sans-serif;}
+
+P,TABLE,TR,TD {font-family:arial,sans-serif;}
+
+UL {list-style-type: square; list-style-image: url(/ceurws-bullet-6x12.png);}
+
+LI {margin-bottom:7px;}   /* separates paper elements */
+
+PRE {color:#228120}
+
+.CEURPAGES {float:right; padding: 2px; font-size:smaller;}
+.AUXPAGES {float:right; padding: 2px; font-size:smaller;}
+
+.CEURTOC {width:90%; page-break-before:always;}    /* table of contents a bit narrower than the total page width */
+
+.CEURAUTHOR {font-style:italic;}
+.CEURAUTHORS {font-style:italic;}
+.AUXAUTHOR {font-style:normal;}
+.AUXAUTHORS {font-style:normal;}
+
+.CEURTITLE {font-weight:normal;}
+.AUXTITLE {font-weight:normal;}
+.CEURTITLE:hover {color: #006040;}   /* proper CEURTITLEs get highlighted */
+
+.CEURVOLNR {font-weight:bold; color:#404040; font-size:large;}
+
+/* write "CEUR-WS.org/" with background color before the VOLNR */
+/* will show up when printing the page */
+/* .CEURVOLNR:before{content:"CEUR-WS.org/"; color:#F5F4EF; font-size:large;} */
+
+.CEURURN {font-weight:bold; color:#404040;}
+
+/* for paper directories */
+.CEURVOLURL {font-weight:bold; color:#000080;}
+.CEURPAPERID {font-weight:bold; color:#404040;}
+
+.CEURABSTRACT {/* border-style: solid; border-width: 1px; padding: 5px 10px 5px 10px; */
+     word-wrap: break-word; margin-left: +42px; margin-right: +68px; }
+
+.unobtrusive {color: #777777; font-size: smaller;} /* unobtrusive text; for copyright notes */
+
+.copyright {font-size: x-small;} /* even smaller text in copyright notes; x-small replaces &lt;font size="-1"&gt; when base size is medium */
+
+
+#CEURWSLOGO {
+   background-image: url('/static/CEUR-WS-logo.png');
+   height: 100px;
+   width: 390px;
+   color-adjust: exact !important;                 /*Firefox*/
+}
+#CEURWSLOGO:hover {
+   background-image: url('/static/CEUR-WS-logo-blue.png'); 
+/*   background-image: url('/static/CEUR-WS-logo-xmas.png'); */
+}
+
+
+#CEURCCBY {
+   background-image: url('/static/cc-by_100x35.png');
+   width: 100px;
+   height: 35px;
+   color-adjust: exact !important;                 /*Firefox*/
+}

tests/test_app.py

@@ -14,8 +14,9 @@ def setUp(self, debug=False, profile=True):
         Basetest.setUp(self, debug=debug, profile=profile)
         script_path=Path(__file__)
         base_path=f"{script_path.parent.parent}/ceur-ws"
+        static_directory=f"{script_path.parent.parent}/static"
         vm=VolumeManager(base_path=base_path)
-        self.ws = WebServer(vm)
+        self.ws = WebServer(vm,static_directory=static_directory)
         self.client = TestClient(self.ws.app)
 
     def checkResponse(self,path:str,status_code:int)->'Response':
@@ -65,7 +66,7 @@ def test_read_volume(self):
         debug=self.debug
         #debug=True
         for ext,expected,equal in [
-            (".json",{'number': 3262.0, 'title':None},True),
+            (".json",{'number': 3262, 'title':None},True),
             ("","CEURVERSION=2020-0",False),
             (".html","CEURVERSION=2020-0",False)
         ]:
@@ -81,4 +82,12 @@ def test_read_volume(self):
                 self.assertEqual(expected,result)
             else:
                 self.assertTrue(expected in result)
+
+    def test_read_paper(self):
+        """
+        test reading the pdf for a paper
+        """
+        response = self.checkResponse(f"/Vol-3262/paper1.pdf",200)
+        self.assertEqual(2509257,response.num_bytes_downloaded)
+        pass
 

tests/test_ceurws.py

@@ -3,17 +3,38 @@
 
 @author: wf
 '''
-from ceurspt.ceurws import Volume
+from ceurspt.ceurws import Volume,VolumeManager
 from tests.basetest import Basetest
+from pathlib import Path
 
 class Test_CEURWS(Basetest):
     """
     Test the ceur-ws concepts
     """
 
+    def setUp(self, debug=False, profile=True):
+        Basetest.setUp(self, debug=debug, profile=profile)
+        script_path=Path(__file__)
+        base_path=f"{script_path.parent.parent}/ceur-ws"
+        self.vm=VolumeManager(base_path=base_path)
+
     def test_volume(self):
         """
         test instantiating a volume
         """
         vol=Volume()
-        print(vol)
+        debug=self.debug
+        if debug:
+            print(vol)
+
+    def test_fix_links(self):
+        """
+        test fixing the links
+        """
+        vol=self.vm.getVolume(3262)
+        html=vol.getHtml()
+        debug=self.debug
+        debug=True
+        if debug:
+            print(html)
+