Many updates:

* add dedicated function to load credentials from a given dictionary, a given file, or a default file
* fix issues with abstracts (multilingual...) and ISSN
* get HAL ID for document when deposit or existing repo
* update example

Author: luclaurent

.gitignore

@@ -161,4 +161,5 @@ cython_debug/
 
 
 ## push2HAL
-.apihal
+.apihal
+push2HAL/version.py

.vscode/settings.json

@@ -0,0 +1,7 @@
+{
+    "python.testing.pytestArgs": [
+        "tests"
+    ],
+    "python.testing.unittestEnabled": false,
+    "python.testing.pytestEnabled": true
+}

README.md

@@ -5,7 +5,7 @@
 
 
 
-`push2HAL` is a basic Python library dedicated to achieving upload on [HAL](https://hal.science) database. It will use the classical API of HAL to get information and the SWORD one to upload content. Two main executables are provided (for UNIX use only):
+`push2HAL` is a basic Python library dedicated to data uploading on [HAL](https://hal.science) database. It will use the classical API of HAL to get information and the SWORD one to upload content. Two main executables are provided (for UNIX use only):
 
 - `pdf2hal` is able to upload a PDF file to an existing notice on HAL (only with valid permission to modify it). 
 - `json2hal` is able to build the necessary data from a JSON file to create a new notice in HAL and upload it directly with or without providing a PDF file.

examples/test.json

@@ -3,6 +3,7 @@
     "lang": "en",
     "title": {"en":"article", "fr":"article"},    
     "subtitle": "A subtitle",
+    "abstract": {"en":"a very long abstract", "fr": "un très long résumé"},
     "authors": [
         {
             "firstname": "John",

examples/test_comments.json

@@ -4,6 +4,7 @@
     "type": "article",      // specific type id see: https://api.archives-ouvertes.fr/ref/doctype (article only works, other types could reuires improvments on code)
     "title": {"en":"article", "fr":"article"},    
     "subtitle": "A subtitle",
+    "abstract": {"en":"a very long abstract", "fr": "un très long résumé"},
     "authors": [
         {
             "firstname": "John",

src/push2HAL/default.py

@@ -66,3 +66,4 @@
 DEFAULT_PEER='0'
 DEFAULT_PROCEEDINGS='0'
 DEFAULT_STRUCT_TYPE='institution'
+DEFAULT_LANG_DOC='en'

src/push2HAL/execHAL.py

@@ -112,7 +112,8 @@ def runJSON2HAL(
 
     # upload to HAL
     if credentials:
-        lib.upload2HAL(file, payload, credentials, server=serverType)
+        id_hal = lib.upload2HAL(file, payload, credentials, server=serverType)
+        return id_hal
     else:
         Logger.error("No provided credentials")
         exitStatus = os.EX_CONFIG

src/push2HAL/libHAL.py

@@ -15,6 +15,7 @@
 import tempfile
 import requests
 import difflib
+import json
 from requests.auth import HTTPBasicAuth
 from lxml import etree
 import re
@@ -324,15 +325,18 @@ def upload2HAL(file, headers, credentials, server="preprod"):
         headers=headers,
         auth=HTTPBasicAuth(credentials["login"], credentials["passwd"]),
     )
-
+    
+     
+    hal_id = None
     if res.status_code == 201:
         Logger.info("Successfully upload to HAL.")
     elif res.status_code == 202:
         Logger.info("Note accepted by HAL.")
         # read return message
         xmlResponse = etree.fromstring(res.text.encode("utf-8"))
         elem = xmlResponse.findall("id", xmlResponse.nsmap)
-        Logger.info("HAL ID: {}".format(elem[0].text))
+        hal_id = elem[0].text
+        Logger.debug("HAL ID: {}".format(elem[0].text))
     elif res.status_code == 401:
         Logger.info("Authentification refused - check credentials")
     else:
@@ -343,14 +347,22 @@ def upload2HAL(file, headers, credentials, server="preprod"):
         )
         Logger.error("Failed to upload. Status code: {}".format(res.status_code))
         if len(elem) > 0:
+            json_ret = list()
             for i in elem:
+                json_ret.append(json.loads(i.text))
                 Logger.warning("Error: {}".format(i.text))
-
+        # extract hal_id
+        for j in json_ret:
+            if j.get('duplicate-entry'):
+                hal_id = list(j.get('duplicate-entry').keys())[0]
+    return hal_id
 
 def setTitles(nInTree, titles, subTitles=None):
     """Add title(s) and subtitle(s) in XML (and specified language)"""
     if subTitles:
         listTitles = {"titles": titles, "subtitles": subTitles}
+    else:
+        listTitles = {"titles": titles}
     nTitle = list()
     for k, n in listTitles.items():
         if type(n) == dict:
@@ -634,19 +646,20 @@ def setAbstract(inTree, abstracts):
     if not abstracts:
         Logger.warning("No provided abstract")
         return None
-    if type(abstracts) != list():
-        abstracts = [abstracts]
     nAbstract = list()
-    for a in abstracts:
-        if type(a) == dict:
-            nAbstract.append(etree.SubElement(inTree, TEI + "abstract"))
-            nAbstract[-1].set(dflt.DEFAULT_XML_LANG + "lang", a["lang"])
-            nAbstract[-1].text = a["text"]
-        else:
-            Logger.warning("No language for abstract: force english")
+    if type(abstracts) == str():
+        Logger.warning("No language for abstract: force english")
+        nAbstract.append(etree.SubElement(inTree, TEI + "abstract"))
+        nAbstract[-1].set(dflt.DEFAULT_XML_LANG + "lang", "en")
+        nAbstract[-1].text = abstracts
+    else:
+        nKeywords = list()
+        for lk, vk in abstracts.items():
             nAbstract.append(etree.SubElement(inTree, TEI + "abstract"))
-            nAbstract[-1].set(dflt.DEFAULT_XML_LANG + "lang", "en")
-            nAbstract[-1].text = a
+            nAbstract[-1].set(dflt.DEFAULT_XML_LANG + "lang", lk)
+            nAbstract[-1].text = vk
+    return nAbstract
+    
     return nAbstract
 
 
@@ -709,7 +722,7 @@ def setIDS(inTree, data):
             Logger.warning("ISSN not valid: {}, continue...".format(data.get("issn")))
     lID.append(setID(inTree, data.get("issn"), "issn"))
     if data.get("eissn", None):
-        if not isbn.validate(data.get("eissn")):
+        if not issn.validate(data.get("eissn")):
             Logger.warning("eISSN not valid: {}, continue...".format(data.get("eissn")))
     lID.append(setID(inTree, data.get("eissn"), "eissn"))
     if data.get("j", None):
@@ -771,8 +784,8 @@ def setLanguage(inTree, language):
     """Set main language in XML"""
     langUsage = etree.SubElement(inTree, TEI + "langUsage")
     if not language:
-        Logger.warning("No language provided - force {}".format(dflt.DEFAULT_XML_LANG))
-        language = dflt.DEFAULT_XML_LANG
+        Logger.warning("No language provided - force {}".format(dflt.DEFAULT_LANG_DOC))
+        language = dflt.DEFAULT_LANG_DOC
     idL = etree.SubElement(langUsage, TEI + "language")
     idL.set("ident", language)
 

src/push2HAL/misc.py

@@ -61,23 +61,46 @@ def showPDFcontent(pdf_path, number=dflt.DEFAULT_NB_CHAR):
         Logger.error(f"Error: {e}")
 
 
-def load_credentials(args):
+def load_credentials(args=None):
     """Load credentials from different sources"""
     cred = dict()
     # if args.hash:
     #     Logger.debug('Load credentials from hash')
     #     cred['hash']=args.hash
-    if args.login and args.passwd:
+    # load input arguments depending on cases
+    login = None
+    passwd = None
+    inputfile = None
+    if args:
+        if type(args) is dict:
+            Logger.debug("Load credentials from dictionary")
+            login = args.get("login",None)
+            passwd = args.get("passwd",None)
+            inputfile = args.get("file",None)
+        elif type(args) is str():
+            Logger.debug("Load credentials from string (file path)")
+            inputfile = args
+        elif type(args) is list():
+            Logger.debug("Load credentials from list (login, passwd)")
+            if len(args)>1:
+                login = args[0]
+                passwd = args[1]
+        else:
+            login = args.login
+            passwd = args.passwd
+            inputfile = args.credentials        
+        
+    if login and passwd:
         Logger.debug("Load credentials from arguments")
         cred["login"] = args.login
         cred["passwd"] = args.passwd
-    elif args.credentials:
-        Logger.debug("Load credentials from file {}".format(args.credentials))
-        if os.path.isfile(args.credentials):
-            with open(args.credentials) as f:
+    elif inputfile:
+        Logger.debug("Load credentials from file {}".format(inputfile))
+        if os.path.isfile(inputfile):
+            with open(inputfile) as f:
                 cred = json.load(f)
     else:
-        Logger.debug("Load credentials from file")
+        Logger.debug("Load credentials from root default file {}".format(dflt.DEFAULT_CREDENTIALS_FILE))
         if os.path.isfile(dflt.DEFAULT_CREDENTIALS_FILE):
             with open(dflt.DEFAULT_CREDENTIALS_FILE) as f:
                 cred = json.load(f)