BibFormat: new HB and HD format templates for the people collection

INSTALL

@@ -83,6 +83,9 @@ Contents
         natively in UTF-8 mode by setting "default-character-set=utf8"
         in various parts of your "my.cnf" file, such as in the
         "[mysql]" part and elsewhere; but this is not really required.
+        Note also that you may encounter problems when MySQL is run in
+        "strict mode"; you may want to configure your "my.cnf" in order
+        to avoid using strict mode (such as `STRICT_ALL_TABLES`).
         <http://mysql.com/>
 
      c) Redis server (may be on a remote machine) for user session

modules/bibdocfile/lib/bibdocfile.py

@@ -59,6 +59,7 @@
 import cgi
 import sys
 import copy
+import tarfile
 
 if sys.hexversion < 0x2060000:
     from md5 import md5
@@ -102,6 +103,7 @@
     encode_for_xml
 from invenio.urlutils import create_url, make_user_agent_string
 from invenio.textutils import nice_size
+from invenio.webuser import collect_user_info
 from invenio.access_control_engine import acc_authorize_action
 from invenio.access_control_admin import acc_is_user_in_role, acc_get_role_id
 from invenio.access_control_firerole import compile_role_definition, acc_firerole_check_user
@@ -123,7 +125,7 @@
     CFG_BIBCATALOG_SYSTEM
 from invenio.bibcatalog import BIBCATALOG_SYSTEM
 from invenio.bibdocfile_config import CFG_BIBDOCFILE_ICON_SUBFORMAT_RE, \
-    CFG_BIBDOCFILE_DEFAULT_ICON_SUBFORMAT
+    CFG_BIBDOCFILE_DEFAULT_ICON_SUBFORMAT, CFG_BIBDOCFILE_STREAM_ARCHIVE_FORMATS
 from invenio.pluginutils import PluginContainer
 
 import invenio.template
@@ -797,17 +799,22 @@ def get_xml_8564(self):
 
         return out
 
-    def get_total_size_latest_version(self):
+    def get_total_size_latest_version(self, user_info=None, subformat=None):
         """
         Returns the total size used on disk by all the files belonging
         to this record and corresponding to the latest version.
 
+        @param user_info: the user_info dictionary, used to check restrictions
+        @type: dict
+        @param subformat: if subformat is specified, it limits files
+            only to those from that specific subformat
+        @type subformat: string
         @return: the total size.
         @rtype: integer
         """
         size = 0
         for (bibdoc, _) in self.bibdocs.values():
-            size += bibdoc.get_total_size_latest_version()
+            size += bibdoc.get_total_size_latest_version(user_info, subformat)
         return size
 
     def get_total_size(self):
@@ -1564,6 +1571,36 @@ def get_text(self, extract_text_if_necessary=True):
 
         return " ".join(texts)
 
+    def stream_archive_of_latest_files(self, req, files_size=''):
+        """
+        Streams the tar archive with all files of a certain file size (that
+        are not restricted or hidden) to the user.
+        File size should be a string that can be compared with the output of
+        BibDocFile.get_subformat() function.
+
+        @param req: Apache Request Object
+        @type req: Apache Request Object
+        @param files_size: size of the files (they can be defined in
+        bibdocfile_config). Empty string means the original size.
+        @type files_size: string
+        """
+        # Get the internal size from the user-friendly file size name
+        internal_format = [f[1] for f in CFG_BIBDOCFILE_STREAM_ARCHIVE_FORMATS if f[0] == files_size]
+        if len(internal_format) < 1:
+            # Incorrect file size
+            return
+        internal_format = internal_format[0]
+        tarname = str(self.id) + "_" + files_size + '.tar'
+
+        # Select files that user can download (not hidden nor restricted)
+        user_info = collect_user_info(req)
+        req.content_type = "application/x-tar"
+        req.headers_out["Content-Disposition"] = 'attachment; filename="%s"' % tarname
+        tar = tarfile.open(fileobj=req, mode='w|')
+        for f in self.list_latest_files():
+            if f.get_subformat() == internal_format and f.is_restricted(user_info)[0] == 0 and not f.hidden:
+                tar.add(f.get_path(), arcname=f.get_full_name(), recursive=False)
+        tar.close()
 
 class BibDoc(object):
     """
@@ -2801,12 +2838,28 @@ def _build_related_file_list(self):
                 cur_doc = BibDoc.create_instance(docid=docid, human_readable=self.human_readable)
                 self.related_files[doctype].append(cur_doc)
 
-    def get_total_size_latest_version(self):
+    def get_total_size_latest_version(self, user_info=None, subformat=None):
         """Return the total size used on disk of all the files belonging
-        to this bibdoc and corresponding to the latest version."""
+        to this bibdoc and corresponding to the latest version. Restricted
+        and hidden files are not counted, unless there is no user_info.
+        @param user_info: the user_info dictionary, used to check restrictions
+        @type: dict
+        @param subformat: if subformat is specified, it limits files
+            only to those from that specific subformat
+        @type subformat: string
+        """
         ret = 0
+        all_files = False
+        # If we are calling this function without user_info, then we want to
+        # see all the files
+        if not user_info:
+            all_files = True
         for bibdocfile in self.list_latest_files():
-            ret += bibdocfile.get_size()
+            # First check for restrictions
+            if all_files or (bibdocfile.is_restricted(user_info)[0] == 0 and not bibdocfile.hidden):
+                # Then check if the format is correct
+                if subformat is None or bibdocfile.get_subformat() == subformat:
+                    ret += bibdocfile.get_size()
         return ret
 
     def get_total_size(self):

modules/bibdocfile/lib/bibdocfile_config.py

@@ -69,3 +69,12 @@
 # CFG_BIBDOCFILE_DEFAULT_ICON_SUBFORMAT -- this is the default subformat used
 # when creating new icons.
 CFG_BIBDOCFILE_DEFAULT_ICON_SUBFORMAT = "icon"
+
+# CFG_BIBDOCFILE_STREAM_ARCHIVE_FORMATS -- a list (not dictionary, because
+# we want to preserve the order) that connects the different format sizes
+# (like 'small', 'medium', etc.) with internal format sizes (like 'icon-180', 'icon-640', etc.)
+CFG_BIBDOCFILE_STREAM_ARCHIVE_FORMATS = [
+    ('small', 'icon-180'),
+    ('medium', 'icon-640'),
+    ('large', 'icon-1440'),
+    ('original', '')]

modules/bibdocfile/lib/bibdocfile_webinterface.py

@@ -83,6 +83,12 @@ def _lookup(self, component, path):
         def getfile(req, form):
             args = wash_urlargd(form, bibdocfile_templates.files_default_urlargd)
             ln = args['ln']
+            if filename[:9] == "allfiles-":
+                files_size = filename[9:]
+                # stream a tar package to the user
+                brd = BibRecDocs(self.recid)
+                brd.stream_archive_of_latest_files(req, files_size)
+                return
 
             _ = gettext_set_language(ln)
 

modules/bibformat/etc/format_templates/Authority_HTML_detailed.bft

@@ -10,8 +10,8 @@
 <BFE_AUTHORITY_JOURNAL detail="yes"/>
 <BFE_AUTHORITY_SUBJECT detail="yes"/>
 
-<BFE_AUTHORITY_AUTHOR_PUBLICATIONS detail="yes"/>
-<BFE_AUTHORITY_LINKS detail="yes"/>
-<BFE_AUTHORITY_CONTROL_NO/>
+<BFE_AUTHORITY_AUTHOR_PUBLICATIONS detail="yes" prefix="<p>" suffix="</p>" />
+<BFE_AUTHORITY_LINKS detail="yes" prefix="<p>" suffix="</p>" />
+<BFE_AUTHORITY_CONTROL_NO prefix="<p>" suffix="</p>" />
 
 </div>

modules/bibformat/etc/format_templates/Institute_HTML_detailed.bft

@@ -7,8 +7,8 @@
 
 <BFE_AUTHORITY_INSTITUTE main_name="no" detail="yes"/>
 
-<BFE_AUTHORITY_PUBLICATIONS detail="yes"/>
-<BFE_AUTHORITY_LINKS detail="yes"/>
-<BFE_AUTHORITY_CONTROL_NO/>
+<BFE_AUTHORITY_PUBLICATIONS detail="yes" prefix="<p>" suffix="</p>" />
+<BFE_AUTHORITY_LINKS detail="yes" prefix="<p>" suffix="</p>" />
+<BFE_AUTHORITY_CONTROL_NO prefix="<p>" suffix="</p>" />
 
 </div>

modules/bibformat/etc/format_templates/People_HTML_brief.bft

@@ -1,4 +1,26 @@
 <name>Default HTML brief</name>
 <description>Brief Authority HTML format.</description>
 
-<BFE_AUTHORITY_AUTHOR detail="no"/>
+<BFE_AUTHORITY_AUTHOR
+    prefix="<span style='font-size: 1.1em;'>"
+    suffix="</span>"
+    detail="no" />
+<BFE_FIELD tag="371__0" prefix="(" suffix=")" />
+
+<small>
+    <ul style="padding: 0; margin: 4px 0 4px 0;">
+        <BFE_AUTHORITY_DESCRIPTION
+            prefix="<li style='display: inline; list-style-type: none;'>"
+            suffix="</li>" />
+        <BFE_AUTHORITY_CONTACT
+            contact_type="email"
+            prefix="<li style='display: inline; list-style-type: none; padding-left: 10px;'>"
+            suffix="</li>"
+            icon="yes" />
+        <BFE_AUTHORITY_CONTACT
+            contact_type="phone"
+            prefix="<li style='display: inline; list-style-type: none; padding-left: 10px;'>"
+            suffix="</li>"
+            icon="yes" />
+    </ul>
+</small>

modules/bibformat/etc/format_templates/People_HTML_detailed.bft

@@ -1,14 +1,132 @@
 <name>Authority HTML detailed</name>
 <description>Detailed Authority HTML format.</description>
 
-<h1><BFE_FIELD tag="100__a"/></h1>
-<style>.inside{margin-left: 10px;}</style>
-<div>
+<style>
+h1 {
+    font-weight: initial;
+    margin: 0 0 10px 0;
+}
+h2 {
+    font-size: initial;
+    font-weight: bold;
+    border-bottom: 2px solid #ccc;
+    padding: 4px;
+    margin: 0 0 10px 0;
+}
+h3 {
+    font-size: initial;
+    font-weight: normal;
+    font-variant: initial;
+    border: none;
+    margin: 0 0 4px 0;
+    padding: 4px;
+    width: auto;
+    border-bottom: 1px solid #ccc;
+    background-color: #eee;
+    color: #555;
+}
+.inside {
+    margin: 10px;
+}
+.row {
+    margin: 0 0 10px 0;
+}
+.row:after {
+    box-sizing: border-box;
+    content: "";
+    clear: both;
+    display: block;
+}
+.column {
+    box-sizing: border-box;
+    float: left;
+    padding: 0 10px 0 10px;
+    width: 33.33%;
+}
+.column ul {
+    margin: 0;
+    padding: 0;
+}
+.column ul li {
+    list-style: none;
+    padding: 4px;
+}
+.column ul li:hover {
+    background-color: #eee;
+}
+ol li {
+    padding: 4px;
+}
+table {
+    border-spacing: 0;
+    margin: 0;
+    width: 100%;
+}
+table tr {
+    display: table-row;
+}
+table tr:hover {
+    background-color: #eee;
+}
+table tr th {
+    text-align: left;
+    padding: 4px;
+}
+table tr td {
+    padding: 4px;
+}
+</style>
 
-<BFE_AUTHORITY_AUTHOR main_name="no" detail="yes"/>
-
-<BFE_AUTHORITY_PUBLICATIONS detail="yes"/>
-<BFE_AUTHORITY_LINKS detail="yes"/>
-<BFE_AUTHORITY_CONTROL_NO/>
+<div class="vcard">
+<div class="row">
+    <BFE_FIELD tag="100__a" prefix="<h1 class='fn'>" suffix="</h1>" />
+    <BFE_FIELD tag="371__0" prefix="<span class='org'>" suffix="</span>" />
+</div>
 
+<h2>Personal Information</h2>
+<div class="row">
+    <div class="column">
+        <h3>Workspace</h3>
+        <table>
+            <tbody>
+                <BFE_FIELD
+                    prefix="<tr><th>Department</th><td>"
+                    suffix="</td></tr>"
+                    tag="371__i" />
+                <BFE_FIELD
+                    prefix="<tr><th>Group</th><td>"
+                    suffix="</td></tr>"
+                    tag="371__g" />
+                <BFE_FIELD
+                    prefix="<tr><th>Section</th><td>"
+                    suffix="</td></tr>"
+                    tag="371__j" />
+                <BFE_AUTHORITY_CONTROL_NO
+                    prefix="<tr><th>Control numbers</th><td>"
+                    suffix="</td></tr>"
+                    print_title="no" />
+          </tbody>
+        </table>
+    </div>
+    <div class="column">
+        <h3>Contact</h3>
+        <BFE_AUTHORITY_CONTACT
+            prefix="<table><tbody>"
+            suffix="</tbody></table>"
+            vcard_support="yes"
+            default="<i>No contact information.</i>" />
+    </div>
+    <div class="column">
+        <h3>Profiles</h3>
+        <BFE_AUTHORITY_LINKS
+            print_title="no"
+            default="<i>No profiles.</i>"/>
+    </div>
 </div>
+</div>
+
+<h2>Publications</h2>
+<BFE_AUTHORITY_PUBLICATIONS
+    print_title="no"
+    detail="yes"
+    default="<i>No publications.</i>" />

modules/bibformat/etc/format_templates/Subject_HTML_detailed.bft

@@ -7,8 +7,8 @@
 
 <BFE_AUTHORITY_SUBJECT detail="yes"/>
 
-<BFE_AUTHORITY_PUBLICATIONS detail="yes"/>
-<BFE_AUTHORITY_LINKS detail="yes"/>
-<BFE_AUTHORITY_CONTROL_NO/>
+<BFE_AUTHORITY_PUBLICATIONS detail="yes" prefix="<p>" suffix="</p>" />
+<BFE_AUTHORITY_LINKS detail="yes" prefix="<p>" suffix="</p>" />
+<BFE_AUTHORITY_CONTROL_NO prefix="<p>" suffix="</p>" />
 
 </div>

modules/bibformat/lib/bibreformat.py

@@ -353,7 +353,11 @@ def task_run_core():
         if task_has_option("last"):
             recids += outdated_caches(fmt, last_updated)
 
-        if task_has_option('ignore_without'):
+        if task_has_option('ignore_without') or \
+                task_has_option('collection') or \
+                task_has_option('field') or \
+                task_has_option('pattern') or \
+                task_has_option('recids'):
             without_fmt = intbitset()
         else:
             without_fmt = missing_caches(fmt)