From 771ba5ae00d0ed641598ab8cb928187f75aff187 Mon Sep 17 00:00:00 2001
From: Christian Boulanger <boulanger@lhlt.mpg.de>
Date: Wed, 31 Jul 2024 21:45:59 +0200
Subject: [PATCH] Add XSLT code
---
convert-anystyle-data/anystyle-to-tei.ipynb | 83 ++++++++++++++++++++-
convert-anystyle-data/lib/.gitignore | 1 +
2 files changed, 81 insertions(+), 3 deletions(-)
create mode 100644 convert-anystyle-data/lib/.gitignore
diff --git a/convert-anystyle-data/anystyle-to-tei.ipynb b/convert-anystyle-data/anystyle-to-tei.ipynb
index f22848a..395c904 100644
--- a/convert-anystyle-data/anystyle-to-tei.ipynb
+++ b/convert-anystyle-data/anystyle-to-tei.ipynb
@@ -148,7 +148,9 @@
{
"metadata": {},
"cell_type": "markdown",
- "source": "## Convert Groundd Truth to TEI",
+ "source": [
+ "## Convert Groundd Truth to TEI"
+ ],
"id": "aaf43ee43bb6d4d"
},
{
@@ -404,19 +406,94 @@
],
"execution_count": 80
},
+ {
+ "cell_type": "markdown",
+ "source": [],
+ "metadata": {
+ "collapsed": false
+ },
+ "id": "8c8b2d820086d461"
+ },
{
"metadata": {},
"cell_type": "markdown",
"source": [
- "## Create LinkML schema from TEI XSD"
+ "## Extract bibliographic data from TEI files "
],
"id": "b0a231dc7bdd8b01"
},
+ {
+ "cell_type": "markdown",
+ "source": [
+ "### Download XSLTs"
+ ],
+ "metadata": {
+ "collapsed": false
+ },
+ "id": "149588c08747c4b3"
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 6,
+ "outputs": [],
+ "source": [
+ "import requests, zipfile, io, os\n",
+ "\n",
+ "if not os.path.isdir('lib/convert'): \n",
+ " url = 'https://github.com/OpenArabicPE/convert_tei-to-bibliographic-data/archive/refs/heads/master.zip'\n",
+ " r = requests.get(url)\n",
+ " assert r.status_code == 200\n",
+ " z = zipfile.ZipFile(io.BytesIO(r.content))\n",
+ " z.extractall('lib')\n",
+ " z.close()\n",
+ " os.rename('lib/convert_tei-to-bibliographic-data-master', 'lib/convert')\n"
+ ],
+ "metadata": {
+ "collapsed": false,
+ "ExecuteTime": {
+ "end_time": "2024-07-31T19:42:48.723119300Z",
+ "start_time": "2024-07-31T19:42:47.234795500Z"
+ }
+ },
+ "id": "1f15b3af6aab73ed"
+ },
{
"metadata": {},
"cell_type": "markdown",
- "source": "",
+ "source": [
+ "### Apply XSLT"
+ ],
"id": "aa86435960e61937"
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "outputs": [],
+ "source": [
+ "from lxml import etree\n",
+ "\n",
+ "def apply_xslt_to_xml(xslt_path, xml_path):\n",
+ " xslt = etree.parse(xslt_path)\n",
+ " xml = etree.parse(xml_path)\n",
+ " transformer = etree.XSLT(xslt)\n",
+ " new_xml = transformer(xml)\n",
+ " return str(new_xml)\n",
+ "\n",
+ "new_xml_str = apply_xslt_to_xml('path_to_your_xslt_file', 'path_to_your_xml_file')\n",
+ "print(new_xml_str)\n"
+ ],
+ "metadata": {
+ "collapsed": false
+ },
+ "id": "cb3b4140ab153c08"
+ },
+ {
+ "cell_type": "markdown",
+ "source": [],
+ "metadata": {
+ "collapsed": false
+ },
+ "id": "387b5b9792505b13"
}
],
"metadata": {
diff --git a/convert-anystyle-data/lib/.gitignore b/convert-anystyle-data/lib/.gitignore
new file mode 100644
index 0000000..f59ec20
--- /dev/null
+++ b/convert-anystyle-data/lib/.gitignore
@@ -0,0 +1 @@
+*
\ No newline at end of file
--
GitLab