diff --git a/common/config.py b/common/config.py
index 370b108732704ae34ddcc3333ed2e569499defea..688a145ad7456e4f531e25d9d8446e89f7ce0745 100644
--- a/common/config.py
+++ b/common/config.py
@@ -16,9 +16,10 @@ class ResourceConfig:
for one of the resources this endpoint communicates with.
"""
+ rxExt = re.compile('\\.[^.]*$')
+
def __init__(self, fnameConfig=None):
self.platform = CorpPlatform.annis
- self.corpus_id = ''
self.transport_protocol = 'https'
self.host = '127.0.0.1'
self.port = '5000'
@@ -168,3 +169,20 @@ class ResourceConfig:
dictConfig = self.processed_gui_config(data)
with open(fnameOut, 'w', encoding='utf-8') as fOut:
json.dump(dictConfig, fOut, sort_keys=True, ensure_ascii=False, indent=2)
+
+
+def read_configs(configDir='./config'):
+ """
+ Load all configuration files from the configuration directory
+ (one file per resource). Initialize ResourceConfig instances for
+ each of them. Return a dictionary where the keys are filenames and
+ the values contain respective ResourceConfig objects.
+ """
+ configs = {}
+ for fname in os.listdir(configDir):
+ if not fname.lower().endswith('.json'):
+ continue
+ fnameFull = os.path.join(configDir, fname)
+ fnameNoExt = ResourceConfig.rxExt.sub('', fname)
+ configs[fnameNoExt] = ResourceConfig(fnameFull)
+ return configs
diff --git a/common/diagnostics.py b/common/diagnostics.py
index e26add3626564df3ee035652b1efd32422f6c224..a9459330e8dff11b01364888b5e2be06837d7b11 100644
--- a/common/diagnostics.py
+++ b/common/diagnostics.py
@@ -12,7 +12,11 @@ class Diagnostic:
"""
fatalFCSDiagnostics = {3, 10, 11} # FCS specifications, 4.2
- fatalSRUDiagnostics = {8, 10} # A subset actually used by this endpoint
+ fatalSRUDiagnostics = {8, 10, 27, 235} # A subset actually used by this endpoint
+
+ stdMessages = {
+ (DiagnosticTypes.sru, 10): 'Something is wrong with the query syntax.'
+ }
def __init__(self, diagType: DiagnosticTypes, diagID: int,
details: str = '',
@@ -24,6 +28,8 @@ class Diagnostic:
self.diagID = diagID
self.details = details
self.message = message
+ if len(self.message) <= 0 and (diagType, diagID) in self.stdMessages:
+ self.message = self.stdMessages[(diagType, diagID)]
self.templateLoader = jinja2.FileSystemLoader(searchpath="./static")
self.templateEnv = jinja2.Environment(loader=self.templateLoader)
diff --git a/common/query_parser.py b/common/query_parser.py
index 32497a480609b6a68fcdcb2ddc609420e3d48792..17affdbaa6204b8f70731ea75de59d711967807a 100644
--- a/common/query_parser.py
+++ b/common/query_parser.py
@@ -1,5 +1,6 @@
from .enums import *
from .diagnostics import Diagnostic
+import re
class QueryParser:
@@ -7,9 +8,39 @@ class QueryParser:
This class contains commonly used methods for initial parsing of a GET
query. It does not include platform-specific methods.
"""
+
+ rxTermQuery = re.compile('^(?:(?:[^ "]|\\\\")*|"(?:[^"]|\\\\")*")$')
+
def __init__(self):
pass
+ @staticmethod
+ def find_operator(strQuery, start=0, end=-1):
+ if end == -1:
+ end = len(strQuery) - 1
+ if strQuery[start:start+3] == 'NOT':
+ return start, 'NOT'
+ parenthBalance = 0
+ inQuotes = False
+ for i in range(start, end):
+ if inQuotes:
+ if strQuery[i] == '"':
+ inQuotes = False
+ continue
+ if strQuery[i] == '"':
+ inQuotes = True
+ continue
+ if strQuery[i] == '(':
+ parenthBalance += 1
+ elif strQuery[i] == ')':
+ parenthBalance -= 1
+ elif parenthBalance == 0:
+ if strQuery[i:i+3] == 'AND':
+ return i, 'AND'
+ elif strQuery[i:i+2] == 'OR':
+ return i, 'OR'
+ return -1, ''
+
def validate_query(self, operation, version, queryType, query,
xFcsEndpointDescription, xFcsContext,
xFcsDataviews, xFcsRewritesAllowed):
diff --git a/common/tsakorpus_query_parser.py b/common/tsakorpus_query_parser.py
new file mode 100644
index 0000000000000000000000000000000000000000..88624a34b21042e5cc774d552c61100fd35c5da2
--- /dev/null
+++ b/common/tsakorpus_query_parser.py
@@ -0,0 +1,73 @@
+import re
+from .query_parser import QueryParser
+from .config import ResourceConfig
+from .diagnostics import Diagnostic, DiagnosticTypes
+
+class TsakorpusQueryParser(QueryParser):
+ """
+ Parses search queries for Tsakorpus-based corpora.
+ """
+
+ def term_query(self, query, config):
+ """
+ Return list of query parameters for one term or sequence of terms.
+ """
+ if len(query) >= 2 and query.startswith('"') and query.endswith('"'):
+ query = query[1:len(query)-1]
+ if len(query) <= 0:
+ return Diagnostic(DiagnosticTypes.sru, 10)
+ getParams = []
+ iTerm = 0
+ for term in query.split(' '):
+ if len(term) > 0:
+ iTerm += 1
+ getParams.append(['w', iTerm, term])
+ if iTerm >= 2:
+ getParams.append(['word_rel_', [iTerm-1, iTerm], '1'])
+ getParams.append(['word_dist_from_', [iTerm-1, iTerm], '1'])
+ getParams.append(['word_dist_to_', [iTerm-1, iTerm], '1'])
+ return getParams
+
+
+ def translate_fcsql(self, query: str, config: ResourceConfig, basicSearch: bool = False, start=0, end=-1):
+ """
+ Translate an FCS-QL query into a Tsakorpus GET query.
+ If something is wrong with the query, return a diagnostic.
+ The function is recursive and only looks at the part of the string
+ delimited by start and end parameters.
+ """
+ if end == -1:
+ end = len(query)
+ if end == 0:
+ return Diagnostic(DiagnosticTypes.sru, 27, message='The query should not be empty.')
+ if self.rxTermQuery.search(query) is not None:
+ return self.term_query(query, config)
+ # if query.count('(') != query.count(')'):
+ # return None
+ if len(query) <= 0:
+ return Diagnostic(DiagnosticTypes.sru, 27, message='The query should not be empty.')
+ if start >= end:
+ return Diagnostic(DiagnosticTypes.sru, 10)
+ iOpPos, strOp = self.find_operator(query, start, end)
+ if iOpPos == -1:
+ if query[start] == '(' and query[end - 1] == ')':
+ return self.translate_fcsql(query, config, basicSearch=basicSearch, start=start + 1, end=end - 1)
+ else:
+ return self.term_query(query[start:end], config)
+ if strOp in ('AND', 'OR'):
+ resultLeft = self.translate_fcsql(query, config, basicSearch=basicSearch, start=start, end=iOpPos)
+ resultRight = self.translate_fcsql(query, config, basicSearch=basicSearch, start=iOpPos + len(strOp), end=end)
+ if (type(resultRight) is Diagnostic or type(resultLeft) is Diagnostic
+ or len(resultLeft) <= 0 or len(resultRight) <= 0):
+ return Diagnostic(DiagnosticTypes.sru, 10)
+ return self.binary_bool(strOp, resultLeft, resultRight, config)
+ elif strOp == 'NOT':
+ resultRight = self.translate_fcsql(query, config, basicSearch=basicSearch, start=iOpPos + len(strOp),
+ end=end)
+ return self.not_bool(resultRight)
+ return {}
+
+
+if __name__ == '__main__':
+ pass
+
diff --git a/config/test.json b/config/test.json
index 1b9ba57aa4866e3a5322d2415e02d69a55ae997b..31ad20d1da233eb4d6edcb686b82f41584ea33dc 100644
--- a/config/test.json
+++ b/config/test.json
@@ -1,5 +1,6 @@
{
"host": "0.0.0.0",
"port": "80",
- "max_hits": 20
+ "max_hits": 20,
+ "platform": "tsakorpus"
}
\ No newline at end of file
diff --git a/main.py b/main.py
index d4b3b2900bcb6995893a5fc11ca47d2bc63dbc83..099c5dad906328f17a613a657cc38e7950cf8fa6 100644
--- a/main.py
+++ b/main.py
@@ -1,33 +1,25 @@
-from fastapi import FastAPI, Request, Query
+from fastapi import FastAPI, Request, Query, Response
from fastapi.staticfiles import StaticFiles
from fastapi.templating import Jinja2Templates
from fastapi.encoders import jsonable_encoder
from fastapi.responses import JSONResponse
from common.query_parser import QueryParser
+from common.tsakorpus_query_parser import TsakorpusQueryParser
from common.enums import *
from common.diagnostics import Diagnostic
-from common.config import ResourceConfig
+from common.config import ResourceConfig, read_configs
import json
import os
import re
import uvicorn
-rxExt = re.compile('\\.[^.]*$')
-
app = FastAPI()
app.mount('/static', StaticFiles(directory='static'), name='static')
templates = Jinja2Templates(directory='static')
app.qp = QueryParser()
-app.configs = {}
-i = 0
-for fname in os.listdir('config'):
- if not fname.lower().endswith('.json'):
- continue
- i += 1
- fnameFull = os.path.join('config', fname)
- fnameNoExt = rxExt.sub('', fname)
- app.configs[fnameNoExt] = ResourceConfig(fnameFull)
+app.qp_tsakorpus = TsakorpusQueryParser()
+app.configs = read_configs()
@app.get('/')
@@ -35,9 +27,8 @@ def root():
return {"message": "Hello World"}
-@app.get('/fcs-endpoint/{platform}/{corpusID}')
+@app.get('/fcs-endpoint/{corpusID}')
def endpoint(
- platform: CorpPlatform,
corpusID: str,
operation: Operation = Operation.explain,
version: SRUVersion = SRUVersion.v2_0,
@@ -60,19 +51,27 @@ def endpoint(
alias='x-fcs-rewrites-allowed'
)
):
+ if corpusID not in app.configs:
+ message = 'No corpus with this ID (' + corpusID +') is served by this Endpoint. ' \
+ 'Valid corpus IDs are: ' + '; '.join(cID for cID in sorted(app.configs)) + '.'
+ diagBody = str(Diagnostic(DiagnosticTypes.sru, 235,
+ message=message)) # "Database does not exist"
+ return Response(content=diagBody, media_type='application/xml')
+ config = app.configs[corpusID]
diagnostics = app.qp.validate_query(operation, version, queryType, query,
xFcsEndpointDescription, xFcsContext,
xFcsDataviews, xFcsRewritesAllowed)
if any(d.is_fatal() for d in diagnostics):
return '\n'.join(str(d) for d in diagnostics)
- if platform == CorpPlatform.annis:
- return {'platform': 'annis', 'operation': operation, 'version': version}
- elif platform == CorpPlatform.litterae:
- return {'platform': 'litterae', 'operation': operation, 'version': version}
- elif platform == CorpPlatform.tsakorpus:
- return {'platform': 'tsakorpus', 'operation': operation, 'version': version}
- return
+ if operation == Operation.searchRetrieve:
+ if config.platform == CorpPlatform.tsakorpus:
+ res = app.qp_tsakorpus.translate_fcsql(query, config)
+ if type(res) == Diagnostic:
+ return Response(content=str(res), media_type='application/xml')
+ return str(res)
+
+ return {'operation': operation, 'version': version}
if __name__ == '__main__':