From 66c1b1f2c3d1a4a4db782e268f9b372b2630b536 Mon Sep 17 00:00:00 2001
From: EsGeh <SamuelGfroerer@googlemail.com>
Date: Wed, 4 Dec 2019 16:15:30 +0100
Subject: [PATCH] tei->html: script creates "self containing" output.

---
 dependencies.conf   |  8 +++++-
 docker-compose.yaml |  1 +
 src/tei2html.py     | 63 +++++++++++++++++++++++++++++++++++++++------
 3 files changed, 63 insertions(+), 9 deletions(-)

diff --git a/dependencies.conf b/dependencies.conf
index 0f6a7fd..3b3e90c 100644
--- a/dependencies.conf
+++ b/dependencies.conf
@@ -13,5 +13,11 @@ hash = 2a01be46ee82fce5eba6074359b3d18db2222e0c
 [eoa-publication-model]
 
 uri = https://github.molgen.mpg.de/EditionOpenAccess/eoa-publication-model.git
-hash = ea8a2eb99b22703f1ef92544352701f07c08d17e
+hash = 5ff326580a6bc34756bce511a23146453fdb82b6
 # init not needed, since only example publication is needed
+
+[webdesign_platform]
+
+uri = https://github.molgen.mpg.de/EditionOpenAccess/webdesign_platform.git
+hash = 7a6c3f7c0db224fdcbb046df3f7edad8fefded3f
+init = ./scripts/init.sh
diff --git a/docker-compose.yaml b/docker-compose.yaml
index 5d9b0a8..7b6319f 100644
--- a/docker-compose.yaml
+++ b/docker-compose.yaml
@@ -12,6 +12,7 @@ services:
       - ./${HOME_DIR}:${HOME_DIR_IN_CONTAINER}
       - ./${INPUT_DIR}:${INPUT_DIR_IN_CONTAINER}
       - ./${OUTPUT_DIR}:${OUTPUT_DIR_IN_CONTAINER}
+      - ./${DEPENDENCIES_DIR}:${DEPENDENCIES_DIR_IN_CONTAINER}
     environment:
       - HOME=${HOME_DIR_IN_CONTAINER}
       - INPUT_DIR=${INPUT_DIR_IN_CONTAINER}
diff --git a/src/tei2html.py b/src/tei2html.py
index 7f9c2ca..a57664a 100755
--- a/src/tei2html.py
+++ b/src/tei2html.py
@@ -15,7 +15,8 @@
 import argparse
 from pathlib import Path
 from os import environ, mkdir
-from shutil import rmtree
+from shutil import rmtree, copytree, ignore_patterns
+from lxml import etree
 
 # things to be done
 # assign ids top to bottom for the following elements:
@@ -39,6 +40,22 @@
 def check_executables():
     check_executable( "saxon" )
 
+def copy_dir(
+        src,
+        dst,
+        **opts
+):
+    print( "'{}' -> '{}'".format( src, dst ) )
+    if Path(dst).exists():
+        rmtree(
+            dst
+        )
+    copytree(
+            src=src,
+            dst=dst,
+            **opts
+    )
+
 if __name__ == '__main__':
     parser = argparse.ArgumentParser(
             formatter_class=argparse.ArgumentDefaultsHelpFormatter
@@ -80,9 +97,9 @@ def check_executables():
     )
     parser.add_argument(
             "-f", "--filename",
-            required = True,
+            default = Path("*.xml"),
             type = Path,
-            help="Name of main EOA-TEI file"
+            help = "xml file inside PUBLICATION_DIR, or absolute path. Patterns like '*.xml' are also acceptable"
     )
     parser.add_argument(
             "-o", "--output-dir",
@@ -94,7 +111,7 @@ def check_executables():
             "-!", "--overwrite",
             action = "store_true",
             default = False,
-            help="overwrite files at OUTPUT_DIR"
+            help="overwrite OUTPUT_DIR, if existing"
     )
     parser.add_argument(
             "--output-file",
@@ -113,6 +130,13 @@ def check_executables():
             type = Path,
             help="static files of publication (relative to OUTPUT_DIR, or absolute)"
     )
+    parser.add_argument(
+            "PUBLICATION_DIR",
+            default = Path("input/example/tei"),
+            help = "directory containing the publication (including resources like pictures, etc.)",
+            nargs = '?', # (optional)
+            type = Path,
+    )
 
     args = parser.parse_args()
 
@@ -130,12 +154,19 @@ def check_executables():
 
     print( "xslt params: " + str(args.param) )
 
-    input_file = args.filename
+    publ_dir = args.PUBLICATION_DIR
+    publ_file = args.filename
+
+    tei_filename = \
+        publ_file if publ_file . is_absolute() else list(publ_dir . glob (str(publ_file)))[0]
+    
+    logging.info( f"tei_file: {tei_filename}, publ_dir: {publ_dir}" )
+
     output_dir = args.output_dir
 
-    if not input_file.is_file():
+    if not tei_filename.is_file():
         raise( Exception(
-            f"not a valid input file: {input_file}"
+            f"not a valid input file: {tei_filename}"
         ) )
 
     if output_dir.exists():
@@ -145,9 +176,25 @@ def check_executables():
             raise( Exception( f"output directory already existing: '{output_dir}'!" ) )
     if not output_dir.exists():
         mkdir( output_dir )
+        ## copy webdesign:
+        copy_dir(
+                DEFAULT_DEPENDENCIES_DIR / "webdesign_platform/dist",
+                output_dir / "webdesign_platform",
+        )
+        ## copy publication static files:
+        static_dest_dir =  output_dir / "publication_static"
+        static_dest_dir.mkdir(
+                parents = True,
+                exist_ok = True
+        )
+        copy_dir(
+                publ_dir,
+                static_dest_dir,
+                ignore = ignore_patterns( "*.xml" )
+        )
 
     run_xslt(
-            input_file = input_file,
+            input_file = tei_filename,
             xslt_file = BASE_DIR / "tei2html" / args.xsl,
             params =
                 [