From 2d5976487027bdbbaf8eeff6a9de9c007827ecc6 Mon Sep 17 00:00:00 2001
From: Christian Boulanger <info@bibliograph.org>
Date: Thu, 22 Aug 2024 11:02:12 +0200
Subject: [PATCH] Download Saxon programmatically
---
convert-anystyle-data/tei-to-bibformats.ipynb | 68 ++++++-------------
1 file changed, 19 insertions(+), 49 deletions(-)
diff --git a/convert-anystyle-data/tei-to-bibformats.ipynb b/convert-anystyle-data/tei-to-bibformats.ipynb
index c448d83..8520080 100644
--- a/convert-anystyle-data/tei-to-bibformats.ipynb
+++ b/convert-anystyle-data/tei-to-bibformats.ipynb
@@ -69,71 +69,41 @@
"metadata": {},
"cell_type": "markdown",
"source": [
- "## Extract bibliographic data from TEI files using XSLT\n",
+ "## Download the Saxon jar\n",
"\n",
- "### Using lxml - currently not working\n"
+ "As the xslt uses v2.0 features, and there are no native-python xslt-2.0 processors, we need to use the Saxon processor\n"
],
- "id": "d08d51f8767602c5"
+ "id": "781d0e0e7a9dd346"
},
{
"metadata": {
"ExecuteTime": {
- "end_time": "2024-08-21T16:24:16.226255Z",
- "start_time": "2024-08-21T16:24:16.196421Z"
+ "end_time": "2024-08-22T08:58:42.538326Z",
+ "start_time": "2024-08-22T08:58:34.687673Z"
}
},
"cell_type": "code",
"source": [
- "from lxml import etree\n",
- "import glob\n",
"import requests\n",
- "\n",
- "def apply_xslt(xslt_path, xml_input_path, xml_output_path):\n",
- " try:\n",
- " xslt_doc = etree.parse(xslt_path)\n",
- " xml_doc = etree.parse(xml_input_path)\n",
- " transformer = etree.XSLT(xslt_doc)\n",
- " new_xml = transformer(xml_doc)\n",
- " with open(xml_output_path, 'w', encoding='utf-8') as f:\n",
- " f.write(new_xml)\n",
- " except etree.XSLTParseError as e:\n",
- " print(f\"Error parsing XSLT file at {xslt_path}: {e}\")\n",
- "\n",
- "for input_path in glob.glob('tei/*.xml'):\n",
- " print(f'Converting {input_path}')\n",
- " base_name = os.path.basename(input_path)\n",
- " output_path = f'tmp/{base_name.replace(\".xml\", \"-mods.xml\")}'\n",
- " apply_xslt('lib/xslt/convert_tei-to-mods_bibl.xsl', input_path, output_path)\n"
- ],
- "id": "af437a5ab3cc41a3",
- "outputs": [
- {
- "name": "stdout",
- "output_type": "stream",
- "text": [
- "Converting tei\\10.1111_1467-6478.00057.xml\n",
- "Error parsing XSLT file at lib/xslt/convert_tei-to-mods_bibl.xsl: xsltParseStylesheetTop: ignoring misplaced import element\n",
- "Converting tei\\10.1111_1467-6478.00080.xml\n",
- "Error parsing XSLT file at lib/xslt/convert_tei-to-mods_bibl.xsl: xsltParseStylesheetTop: ignoring misplaced import element\n",
- "Converting tei\\10.1515_zfrs-1980-0103.xml\n",
- "Error parsing XSLT file at lib/xslt/convert_tei-to-mods_bibl.xsl: xsltParseStylesheetTop: ignoring misplaced import element\n",
- "Converting tei\\10.1515_zfrs-1980-0104.xml\n",
- "Error parsing XSLT file at lib/xslt/convert_tei-to-mods_bibl.xsl: xsltParseStylesheetTop: ignoring misplaced import element\n"
- ]
- }
+ "import zipfile\n",
+ "import io\n",
+ "import os\n",
+ "url = \"https://github.com/Saxonica/Saxon-HE/releases/download/SaxonHE12-5/SaxonHE12-5J.zip\"\n",
+ "target_dir = 'lib/SaxonHE12-5'\n",
+ "response = requests.get(url, stream=True)\n",
+ "file_zip = zipfile.ZipFile(io.BytesIO(response.content))\n",
+ "os.makedirs(target_dir, exist_ok=True)\n",
+ "file_zip.extractall(path=target_dir)"
],
- "execution_count": 41
+ "id": "72b688e9b2e0d1f2",
+ "outputs": [],
+ "execution_count": 86
},
{
"metadata": {},
"cell_type": "markdown",
- "source": [
- "### Using Saxon:\n",
- "\n",
- "- download ZIP from https://github.com/Saxonica/Saxon-HE/releases/download/SaxonHE12-5/SaxonHE12-5J.zip\n",
- "- unpack in lib/SaxonHE12-5J"
- ],
- "id": "781d0e0e7a9dd346"
+ "source": "## Run the transformation",
+ "id": "1bbb36ac0f4fd1b5"
},
{
"metadata": {
--
GitLab