From 7e3e5502f18822abece61afa374fc158f9b3d119 Mon Sep 17 00:00:00 2001 From: filak Date: Fri, 6 Oct 2023 12:45:25 +0200 Subject: [PATCH] Version 1.5.3 --- flask-app/application/__init__.py | 2 +- flask-app/application/routes.py | 10 +- flask-app/application/templates/manage.html | 29 +++--- flask-app/application/utils.py | 103 ++++++++++---------- flask-app/mtw_requirements.txt | 2 +- 5 files changed, 73 insertions(+), 73 deletions(-) diff --git a/flask-app/application/__init__.py b/flask-app/application/__init__.py index 0c089d2..2c27fa1 100644 --- a/flask-app/application/__init__.py +++ b/flask-app/application/__init__.py @@ -106,7 +106,7 @@ def getPath(path): ## SeaSurf (csrf) & Talisman csrf.init_app(app) - if not relax: + if not relax and not app.debug: ## Paranoid paranoid.init_app(app) paranoid.redirect_view = getPath('/') diff --git a/flask-app/application/routes.py b/flask-app/application/routes.py index 0152d2a..4834990 100644 --- a/flask-app/application/routes.py +++ b/flask-app/application/routes.py @@ -1385,6 +1385,13 @@ def manage(action): exp_files[f]['fname'] = fp.name exp_files[f]['fdate'] = mtu.getFpathDate(fp) + + # 2024_umls.tsv + show_umls_exports = False + umls_tsv = mtu.getTempFpath('umls', ext='tsv') + if umls_tsv.is_file(): + show_umls_exports = True + show_lookups_exports = False lookups_base = mtu.getTempFpath('lookups') if lookups_base.is_file(): @@ -1400,7 +1407,8 @@ def manage(action): msg = 'Background worker is RUNNING...' flash(msg, 'info') - return render_template('manage.html', users=users, exports=exp_files, show_lookups=show_lookups_exports, show_marc=show_marc_exports) + return render_template('manage.html', users=users, exports=exp_files, + show_lookups=show_lookups_exports, show_marc=show_marc_exports, show_umls_exports=show_umls_exports) @app.route(getPath('/download/')) diff --git a/flask-app/application/templates/manage.html b/flask-app/application/templates/manage.html index 3a94da6..a94779d 100644 --- a/flask-app/application/templates/manage.html +++ b/flask-app/application/templates/manage.html @@ -17,9 +17,7 @@
- {% include 'snips/users.html' %} -
@@ -232,7 +230,7 @@

MARC21 Exports

-{% if show_marc %} + {% if show_marc %}
@@ -341,7 +339,7 @@
Settings
- {% endif %} + {% endif %}
@@ -484,7 +482,6 @@
JSON for data parsers
-
JSON for Elastic
@@ -518,14 +515,16 @@
JSON for Elastic
-
UMLS TSV for MSSQL - includes inactive terms
+
UMLS TSV MSSQL [includes inactive terms]
- {% if show_marc %} + {% if show_umls_exports %}
{% if exports['umls_all_tsv'] %} - {% else %} @@ -549,14 +548,16 @@
UMLS TSV for MSSQL - includes inactive terms
-
UMLS TSV RAW Includes inactive terms; chars NOT normalized
+
UMLS TSV RAW [includes inactive terms; chars NOT normalized]
- {% if show_marc %} + {% if show_umls_exports %} {% if exports['umls_raw_tsv'] %} - {% else %} @@ -578,12 +579,8 @@
UMLS TSV RAW Includes inactive terms; chars NOT nor
- - {% endif %} - + {% endif %}
- -
diff --git a/flask-app/application/utils.py b/flask-app/application/utils.py index ea02962..4f9a890 100644 --- a/flask-app/application/utils.py +++ b/flask-app/application/utils.py @@ -156,7 +156,7 @@ def exportData(export): if export in ['umls_all','umls_raw']: fpath = getTempFpath('umls', ext=ext) epath = getStatsFpath(export, ext=ext) - exportTsv(export, fpath, epath) + exportTsvFile(export, fpath, epath) else: fpath = getTempFpath(export, ext=ext) if not lpath.is_file(): @@ -164,7 +164,7 @@ def exportData(export): if resp and fpath.is_file(): epath = getStatsFpath(export, ext=ext) - exportTsv(export, fpath, epath) + exportTsvFile(export, fpath, epath) def startStats(stat, fpath, lpath, interval=None): @@ -1210,6 +1210,12 @@ def writeOutputGzip(fpath, data, mode='wt'): ft.write(data) +def writeOutputGzipTsv(fpath, lines, mode='wt'): + with gzip.open(str(fpath)+'.gz', mode=mode, encoding='utf-8', newline='') as ft: + writer = csv.writer(ft, delimiter='\t', doublequote=True, quoting=1) + writer.writerows(lines) + + def loadJsonFile(fpath, default=None): try: with open(str(fpath), mode='r', encoding='utf-8') as json_file: @@ -1236,7 +1242,7 @@ def sanitize_input(text, normalize=True): t = ' '.join(t.split()) t = t.replace('?','') if normalize: - t = normalize_str(t) + t = normalize_str(t, clear_escaping=True) t = t.replace('"','\\"') return t @@ -1249,13 +1255,13 @@ def sanitize_text_query(text): return t -def normalize_str(text, escape_double_chars_tsv=False, skip_normalization=False): +def normalize_str(text, skip_normalization=False, clear_escaping=False): if text: if not skip_normalization: for src, trg in app.config['CHAR_NORM_MAP']: text = text.replace(src, trg) - if escape_double_chars_tsv: - text = text.replace('\\"','""') + if clear_escaping: + text = text.replace('\\"','"') return text @@ -1531,81 +1537,70 @@ def getDescClass(dtype): return dtype_dict.get(dtype, '-1') -def exportTsv(export, inputFile, outputFile): - - count = 0 - found = 0 - batch = 0 - - bsize = 30000 +def exportTsvFile(export, inputFile, outputFile): ext = os.path.splitext(inputFile)[1] fext = ext.lower() - - head = '' tab = '\t' if export == 'umls': ### ?dui ?cui ?lang ?tty ?str ?tui ?scn cols = 'DescriptorUI,ConceptUI,Language,TermType,String,TermUI,ScopeNote'.split(',') - head = (tab).join(cols) + '\n' + elif export in ['umls_all','umls_raw']: ### ?status ?tstatus ?dui ?cui ?lang ?tty ?str ?tui ?scn cols = 'Dstatus,Tstatus,DescriptorUI,ConceptUI,Language,TermType,String,TermUI,ScopeNote'.split(',') - head = (tab).join(cols) + '\n' - writeOutputGzip(outputFile, head) + writeOutputGzipTsv(outputFile, [cols], mode='wt') if fext == '.gz': fh = gzip.open(str(inputFile), mode='rt', encoding='utf-8') else: fh = open(str(inputFile), mode='r', encoding='utf-8') - s = io.StringIO() + docs = [] + count = 0 for line in fh: - count += 1 - found += 1 - batch += 1 - - if count > 1: + if count > 0: line = line.replace('@'+app.config['TARGET_LANG'], '') line = line.replace('^^xsd:boolean', '') - row = line.split(tab) + + row = line.strip().split(tab) if export in ['umls_all','umls_raw']: if row[1] == 'false': #print(line) pass else: - line = (tab).join(row) + clean_row = (tab).join( clearQuotes(row) ) + if export == 'umls_raw': - s.write( normalize_str(line, escape_double_chars_tsv=True, skip_normalization=True) ) + docs.append( normalize_str( clean_row, clear_escaping=True, skip_normalization=True ).split(tab) ) else: - s.write( normalize_str(line, escape_double_chars_tsv=True) ) + docs.append( normalize_str( clean_row, clear_escaping=True ).split(tab) ) elif export == 'umls': - if row[0] == 'false' or row[1] == 'false': #print(line) pass else: - line = (tab).join(row[2:]) - s.write( normalize_str(line, escape_double_chars_tsv=True) ) + clean_row = (tab).join( clearQuotes(row[2:]) ) + docs.append( normalize_str( clean_row, clear_escaping=True ).split(tab) ) - else: - line = (tab).join(row) - s.write(line) + count += 1 + + writeOutputGzipTsv(outputFile, docs, mode='at') - if batch == bsize: - writeOutputGzip(outputFile, s.getvalue(), mode='at') - batch = 0 - s.close() - s = io.StringIO() - fh.close() - writeOutputGzip(outputFile, s.getvalue(), mode='at') - s.close() +def clearQuotes(row): + cleaned = [] + for item in row: + if item.startswith('"'): + cleaned.append(item[1:-1]) + else: + cleaned.append(item) + return cleaned def cleanDescView(dview): @@ -1627,18 +1622,18 @@ def cleanDescView(dview): line = line.strip() if line.startswith('type:'): - if 'MeSH TopicalDescriptor' in line: - pass - elif 'MeSH GeographicalDescriptor' in line: - pass - elif 'MeSH PublicationType' in line: - pass - elif 'MeSH CheckTag' in line: - pass - elif 'MeSH Qualifier' in line: - pass - elif line: - out += line + '\n' + if 'MeSH TopicalDescriptor' in line: + pass + elif 'MeSH GeographicalDescriptor' in line: + pass + elif 'MeSH PublicationType' in line: + pass + elif 'MeSH CheckTag' in line: + pass + elif 'MeSH Qualifier' in line: + pass + elif line: + out += line + '\n' else: if line: out += line + '\n' diff --git a/flask-app/mtw_requirements.txt b/flask-app/mtw_requirements.txt index 9c4fbc3..04ac62f 100644 --- a/flask-app/mtw_requirements.txt +++ b/flask-app/mtw_requirements.txt @@ -1,4 +1,4 @@ -### Version 1.5.2 +### Version 1.5.3 ### Python 3.11 ### arrow==1.3.0 bcrypt==4.0.1 ## NOT python-bcrypt !!!