Module tf.browser.ner.serve
Main controller for Flask
This module contains the controllers that Flask invokes when serving the annotation tool in the TF browser.
Expand source code Browse git
"""Main controller for Flask
This module contains the controllers that Flask invokes when serving
the annotation tool in the TF browser.
"""
from flask import render_template
from ...ner.settings import TOOLKEY
from .websettings import SC_ALL
from .request import Request
from .fragments import Fragments
class Serve(Request, Fragments):
def __init__(self, web):
"""Object that implements the controller functions for the annotation tool.
Parameters
----------
web: object
This represents the Flask website that is the TF browser.
It has initialized a `tf.ner.ner.NER` object,
and has stored it under attribute `ner`.
See `tf.browser.ner.web.factory` and `tf.browser.web.factory`.
"""
self.web = web
ner = web.ner
self.ner = ner
Request.__init__(self)
self.initVars()
def setupFull(self):
"""Prepares to serve a complete page.
* Sets up the find widget;
* Encodes the active entity in hidden `input` elements;
* Collects and generates the specific CSS styles needed for this corpus.
"""
ner = self.ner
v = self.v
self.findSetup()
self.wrapActive()
v.css = ner.css
def setupLean(self):
"""Prepares to update a portion of the page.
* Encodes the active entity in hidden `input` elements.
"""
self.wrapActive()
def actionsFull(self):
"""Carries out requested actions before building the full page.
* annotation set management actions;
* fetch selected buckets from the whole corpus;
* modification actions in the selected set.
"""
self.setHandling()
self.getBuckets()
self.updateHandling()
def actionsLean(self, node):
"""Carries out requested actions before building a portion of the page.
* fetch all buckets from a section of the corpus.
"""
self.getBuckets(node=node)
def wrapFull(self):
"""Builds the full page.
This includes the controls by which the user makes selections and triggers
actions.
"""
ner = self.ner
v = self.v
sortKey = v.sortkey
sortDir = v.sortdir
subtleFilter = v.subtlefilter
activeEntity = v.activeentity
activeTrigger = v.activetrigger
tokenStart = v.tokenstart
tokenEnd = v.tokenend
excludedTokens = v.excludedtokens
buckets = self.buckets
self.wrapQuery()
v.entitytable = (
ner.showEntities(
activeEntity=activeEntity, sortKey=sortKey, sortDir=sortDir
)
if ner.sheetName is None
else ner.showTriggers(
activeEntity=activeEntity,
activeTrigger=activeTrigger,
sortKey=sortKey,
sortDir=sortDir,
subtleFilter=subtleFilter,
)
)
v.entityoverview = ner.showEntityOverview()
v.entityheaders = self.wrapEntityHeaders()
v.buckets = ner.showContent(
buckets,
activeEntity=activeEntity,
activeTrigger=activeTrigger,
excludedTokens=excludedTokens,
mayLimit=not (tokenStart and tokenEnd),
)
return render_template(f"{TOOLKEY}/index.html", **v)
def wrapLean(self):
"""Builds a portion of the page.
No need to build user controls, because they are already on the page.
Returns
-------
The generated HTML for the portion of the page.
"""
ner = self.ner
v = self.v
activeEntity = v.activeentity
activeTrigger = v.activetrigger
excludedTokens = v.excludedtokens
buckets = self.buckets
return ner.showContent(
buckets,
activeEntity=activeEntity,
activeTrigger=activeTrigger,
excludedTokens=excludedTokens,
mayLimit=False,
)
def getBuckets(self, noFind=False, node=None):
"""Fetch a selection of buckets from the corpus.
The selection is defined in the `v`.
We further modify the selection by two additional parameters.
The resulting list of buckets is obtained by
`tf.ner.corpus.Corpus.filterContent`, and each member in the bucket
list is a tuple as indicated in the `filterContent` function.
The list is stored in the `Serve` object.
Additionally, statistics about these buckets and how many entity values
occur in it, are delivered in the `v`.
Parameters
----------
noFind: boolean, optional False
If `noFind` we override the filtering by the filter widget on the interface.
We use this when the user has indicated that he wants to apply an action
on all buckets instead of the filtered ones.
node: integer, optional None
If passed, it is a TF node, probably for a top-level section.
The effect is that it restricts the result to those buttons that fall
under that TF node.
We use this when we retrieve the context for a given bucket.
"""
ner = self.ner
v = self.v
bFindRe = None if noFind else v.bfindre
anyEnt = v.anyent
activeEntity = v.activeentity
activeTrigger = v.activetrigger
tokenStart = v.tokenstart
tokenEnd = v.tokenend
valSelect = v.valselect
freeState = v.freestate
setData = ner.getSetData()
entityIdent = setData.entityIdent
if activeEntity not in entityIdent:
activeEntity = None
v.activeentity = None
v.activetrigger = None
if activeTrigger is None:
activeTrigger = None
v.activetrigger = None
qTokens = (
ner.stringsFromTokens(tokenStart, tokenEnd)
if tokenStart and tokenEnd
else None
)
(
self.buckets,
v.nfind,
v.nvisible,
v.nent,
) = ner.filterContent(
node=node,
bFindRe=bFindRe,
anyEnt=anyEnt,
eVals=activeEntity,
trigger=activeTrigger,
qTokens=qTokens,
valSelect=valSelect,
freeState=freeState,
)
def setHandling(self):
"""Carries out the set-related actions before composing the page.
These actions are:
* read the list of sets again
* switch to an other set than the current set
and create it if it does not yet exist;
* duplicate the current set;
* rename the current set;
* delete a set.
The results of the actions are wrapped in messages and stored in the
`v`.
"""
ner = self.ner
ner.readSets()
ner.readSheets()
v = self.v
chosenTask = v.task
dupSet = v.duset
renamedSet = v.rset
deleteSet = v.dset
messages = []
if deleteSet:
messages.extend(ner.setDel(deleteSet))
v.dset = ""
v.task = ""
if dupSet:
messages.extend(ner.setDup(dupSet))
v.task = dupSet
v.duset = ""
if renamedSet and chosenTask:
messages.extend(ner.setMove(renamedSet))
v.task = renamedSet
v.rset = ""
v.messagesrc = messages
chosenTask = v.task
sheetCase = v.sheetcase
ner.setTask(chosenTask, caseSensitive=sheetCase)
ner.loadSetData()
self.wrapSets()
self.wrapMessages()
self.wrapCaption()
self.wrapLogs()
def updateHandling(self):
"""Carries out modification actions in the current annotation set.
Modification actions are:
* deletion of an entity;
* addition of an entity.
The results of the actions are wrapped in a report and stored in the
`v`.
"""
ner = self.ner
v = self.v
delData = v.deldata
addData = v.adddata
activeEntity = v.activeentity
tokenStart = v.tokenstart
tokenEnd = v.tokenend
submitter = v.submitter
excludedTokens = v.excludedtokens
bFindRe = v.bfindre
scope = v.scope
hasEnt = activeEntity is not None
hasOcc = tokenStart is not None and tokenEnd is not None
if (
submitter in {"delgo", "addgo"}
and (delData or addData)
and (hasEnt or hasOcc)
):
if bFindRe and scope == SC_ALL:
self.getBuckets(noFind=True)
if submitter == "delgo" and delData:
report = ner.delEntityRich(
delData.deletions, self.buckets, excludedTokens=excludedTokens
)
ner.loadSetData()
self.wrapReport(report, "del")
if hasEnt:
setData = ner.getSetData()
entityIdent = setData.entityIdent
stillExists = activeEntity in entityIdent
if not stillExists:
v.activeentity = None
v.activetrigger = None
if submitter == "addgo" and addData:
report = ner.addEntityRich(
addData.additions, self.buckets, excludedTokens=excludedTokens
)
ner.loadSetData()
self.wrapReport(report, "add")
self.adaptValSelect()
self.getBuckets()
def serveNer(web):
"""Main controller to render a full page.
Parameters
----------
web: object
The TF browser object, a Flask web app.
"""
serve = Serve(web)
serve.setupFull()
serve.actionsFull()
return serve.wrapFull()
def serveNerContext(web, node):
"""Controller to render a portion of a page.
More specifically: the context around a single bucket.
Parameters
----------
web: object
The TF browser object, a Flask web app.
node: integer
The TF node that contain the bucket nodes that form the context.
"""
serve = Serve(web)
serve.setupLean()
serve.actionsLean(node)
return serve.wrapLean()
Functions
def serveNer(web)
-
Main controller to render a full page.
Parameters
web
:object
- The TF browser object, a Flask web app.
def serveNerContext(web, node)
-
Controller to render a portion of a page.
More specifically: the context around a single bucket.
Parameters
web
:object
- The TF browser object, a Flask web app.
node
:integer
- The TF node that contain the bucket nodes that form the context.
Classes
class Serve (web)
-
Object that implements the controller functions for the annotation tool.
Parameters
Expand source code Browse git
class Serve(Request, Fragments): def __init__(self, web): """Object that implements the controller functions for the annotation tool. Parameters ---------- web: object This represents the Flask website that is the TF browser. It has initialized a `tf.ner.ner.NER` object, and has stored it under attribute `ner`. See `tf.browser.ner.web.factory` and `tf.browser.web.factory`. """ self.web = web ner = web.ner self.ner = ner Request.__init__(self) self.initVars() def setupFull(self): """Prepares to serve a complete page. * Sets up the find widget; * Encodes the active entity in hidden `input` elements; * Collects and generates the specific CSS styles needed for this corpus. """ ner = self.ner v = self.v self.findSetup() self.wrapActive() v.css = ner.css def setupLean(self): """Prepares to update a portion of the page. * Encodes the active entity in hidden `input` elements. """ self.wrapActive() def actionsFull(self): """Carries out requested actions before building the full page. * annotation set management actions; * fetch selected buckets from the whole corpus; * modification actions in the selected set. """ self.setHandling() self.getBuckets() self.updateHandling() def actionsLean(self, node): """Carries out requested actions before building a portion of the page. * fetch all buckets from a section of the corpus. """ self.getBuckets(node=node) def wrapFull(self): """Builds the full page. This includes the controls by which the user makes selections and triggers actions. """ ner = self.ner v = self.v sortKey = v.sortkey sortDir = v.sortdir subtleFilter = v.subtlefilter activeEntity = v.activeentity activeTrigger = v.activetrigger tokenStart = v.tokenstart tokenEnd = v.tokenend excludedTokens = v.excludedtokens buckets = self.buckets self.wrapQuery() v.entitytable = ( ner.showEntities( activeEntity=activeEntity, sortKey=sortKey, sortDir=sortDir ) if ner.sheetName is None else ner.showTriggers( activeEntity=activeEntity, activeTrigger=activeTrigger, sortKey=sortKey, sortDir=sortDir, subtleFilter=subtleFilter, ) ) v.entityoverview = ner.showEntityOverview() v.entityheaders = self.wrapEntityHeaders() v.buckets = ner.showContent( buckets, activeEntity=activeEntity, activeTrigger=activeTrigger, excludedTokens=excludedTokens, mayLimit=not (tokenStart and tokenEnd), ) return render_template(f"{TOOLKEY}/index.html", **v) def wrapLean(self): """Builds a portion of the page. No need to build user controls, because they are already on the page. Returns ------- The generated HTML for the portion of the page. """ ner = self.ner v = self.v activeEntity = v.activeentity activeTrigger = v.activetrigger excludedTokens = v.excludedtokens buckets = self.buckets return ner.showContent( buckets, activeEntity=activeEntity, activeTrigger=activeTrigger, excludedTokens=excludedTokens, mayLimit=False, ) def getBuckets(self, noFind=False, node=None): """Fetch a selection of buckets from the corpus. The selection is defined in the `v`. We further modify the selection by two additional parameters. The resulting list of buckets is obtained by `tf.ner.corpus.Corpus.filterContent`, and each member in the bucket list is a tuple as indicated in the `filterContent` function. The list is stored in the `Serve` object. Additionally, statistics about these buckets and how many entity values occur in it, are delivered in the `v`. Parameters ---------- noFind: boolean, optional False If `noFind` we override the filtering by the filter widget on the interface. We use this when the user has indicated that he wants to apply an action on all buckets instead of the filtered ones. node: integer, optional None If passed, it is a TF node, probably for a top-level section. The effect is that it restricts the result to those buttons that fall under that TF node. We use this when we retrieve the context for a given bucket. """ ner = self.ner v = self.v bFindRe = None if noFind else v.bfindre anyEnt = v.anyent activeEntity = v.activeentity activeTrigger = v.activetrigger tokenStart = v.tokenstart tokenEnd = v.tokenend valSelect = v.valselect freeState = v.freestate setData = ner.getSetData() entityIdent = setData.entityIdent if activeEntity not in entityIdent: activeEntity = None v.activeentity = None v.activetrigger = None if activeTrigger is None: activeTrigger = None v.activetrigger = None qTokens = ( ner.stringsFromTokens(tokenStart, tokenEnd) if tokenStart and tokenEnd else None ) ( self.buckets, v.nfind, v.nvisible, v.nent, ) = ner.filterContent( node=node, bFindRe=bFindRe, anyEnt=anyEnt, eVals=activeEntity, trigger=activeTrigger, qTokens=qTokens, valSelect=valSelect, freeState=freeState, ) def setHandling(self): """Carries out the set-related actions before composing the page. These actions are: * read the list of sets again * switch to an other set than the current set and create it if it does not yet exist; * duplicate the current set; * rename the current set; * delete a set. The results of the actions are wrapped in messages and stored in the `v`. """ ner = self.ner ner.readSets() ner.readSheets() v = self.v chosenTask = v.task dupSet = v.duset renamedSet = v.rset deleteSet = v.dset messages = [] if deleteSet: messages.extend(ner.setDel(deleteSet)) v.dset = "" v.task = "" if dupSet: messages.extend(ner.setDup(dupSet)) v.task = dupSet v.duset = "" if renamedSet and chosenTask: messages.extend(ner.setMove(renamedSet)) v.task = renamedSet v.rset = "" v.messagesrc = messages chosenTask = v.task sheetCase = v.sheetcase ner.setTask(chosenTask, caseSensitive=sheetCase) ner.loadSetData() self.wrapSets() self.wrapMessages() self.wrapCaption() self.wrapLogs() def updateHandling(self): """Carries out modification actions in the current annotation set. Modification actions are: * deletion of an entity; * addition of an entity. The results of the actions are wrapped in a report and stored in the `v`. """ ner = self.ner v = self.v delData = v.deldata addData = v.adddata activeEntity = v.activeentity tokenStart = v.tokenstart tokenEnd = v.tokenend submitter = v.submitter excludedTokens = v.excludedtokens bFindRe = v.bfindre scope = v.scope hasEnt = activeEntity is not None hasOcc = tokenStart is not None and tokenEnd is not None if ( submitter in {"delgo", "addgo"} and (delData or addData) and (hasEnt or hasOcc) ): if bFindRe and scope == SC_ALL: self.getBuckets(noFind=True) if submitter == "delgo" and delData: report = ner.delEntityRich( delData.deletions, self.buckets, excludedTokens=excludedTokens ) ner.loadSetData() self.wrapReport(report, "del") if hasEnt: setData = ner.getSetData() entityIdent = setData.entityIdent stillExists = activeEntity in entityIdent if not stillExists: v.activeentity = None v.activetrigger = None if submitter == "addgo" and addData: report = ner.addEntityRich( addData.additions, self.buckets, excludedTokens=excludedTokens ) ner.loadSetData() self.wrapReport(report, "add") self.adaptValSelect() self.getBuckets()
Ancestors
Methods
def actionsFull(self)
-
Carries out requested actions before building the full page.
- annotation set management actions;
- fetch selected buckets from the whole corpus;
- modification actions in the selected set.
def actionsLean(self, node)
-
Carries out requested actions before building a portion of the page.
- fetch all buckets from a section of the corpus.
def getBuckets(self, noFind=False, node=None)
-
Fetch a selection of buckets from the corpus.
The selection is defined in the
v
.We further modify the selection by two additional parameters.
The resulting list of buckets is obtained by
Corpus.filterContent()
, and each member in the bucket list is a tuple as indicated in thefilterContent
function. The list is stored in theServe
object. Additionally, statistics about these buckets and how many entity values occur in it, are delivered in thev
.Parameters
noFind
:boolean
, optionalFalse
-
If
noFind
we override the filtering by the filter widget on the interface.We use this when the user has indicated that he wants to apply an action on all buckets instead of the filtered ones.
node
:integer
, optionalNone
-
If passed, it is a TF node, probably for a top-level section. The effect is that it restricts the result to those buttons that fall under that TF node.
We use this when we retrieve the context for a given bucket.
def setHandling(self)
-
Carries out the set-related actions before composing the page.
These actions are:
- read the list of sets again
- switch to an other set than the current set and create it if it does not yet exist;
- duplicate the current set;
- rename the current set;
- delete a set.
The results of the actions are wrapped in messages and stored in the
v
. def setupFull(self)
-
Prepares to serve a complete page.
- Sets up the find widget;
- Encodes the active entity in hidden
input
elements; - Collects and generates the specific CSS styles needed for this corpus.
def setupLean(self)
-
Prepares to update a portion of the page.
- Encodes the active entity in hidden
input
elements.
- Encodes the active entity in hidden
def updateHandling(self)
-
Carries out modification actions in the current annotation set.
Modification actions are:
- deletion of an entity;
- addition of an entity.
The results of the actions are wrapped in a report and stored in the
v
. def wrapFull(self)
-
Builds the full page.
This includes the controls by which the user makes selections and triggers actions.
def wrapLean(self)
-
Builds a portion of the page.
No need to build user controls, because they are already on the page.
Returns
The generated HTML for the portion of the page.
Inherited members