o h=@sddlZddlZddlZddlmmZddlmm m Z ddl Z ddl mZddl mZddlTddlmZddlTddlmZddlmZddlZddlmZmZmZddlZd eeeffd d Zd eefd d Zd ee e!e!ffddZ"d e fddZ#d eeeeeee e!e!e fffddZ$ d(dededeede de dede!de!de de d dfdd Z%d!d"Z&d#d$Z'd%d&Z(e)d'kre(dSdS))N)visualize_search_str)edit_html_overflow)*)CoNLL)Pipeline)StringIO)ListTupleAnyreturncCs(tjdttd}tjdttd}||fS)z Gets user input for the Semgrex text and queries to process. @return: A tuple containing the user's input text and their input queries Text to analyze placeholder9Semgrex search queries (separate each query with a comma))st text_areaDEFAULT_SAMPLE_TEXTDEFAULT_SEMGREX_QUERY) input_txt input_queriesra/var/www/html/env_mimamsha/lib/python3.10/site-packages/stanza/utils/visualization/semgrex_app.pyget_semgrex_text_and_querysrcCsPtdtjdddd}g}|D]}t|d}|}||q|S)z Allows user to submit files for analysis. @return: List of strings containing the file contents of each submitted file. The i-th element of res is the string representing the i-th file uploaded. z-**Alternatively, upload file(s) to analyze.** button_labelT collapsedaccept_multiple_fileslabel_visibilityutf-8)rmarkdown file_uploaderrgetvaluedecodereadappend)uploaded_filesresfilestringio string_datarrrget_file_input(s  r*cCsJtjddd}d\}}|r tjdddd}tjd d d |d d }|||fS)a9 Allows user to specify a specific window of Semgrex hits to visualize. Works similar to Python splicing. @return: A tuple containing a bool representing whether or not the user wants to visualize a splice of the visualizations, and two ints representing the start and end indices of the splice. z3Visualize a specific window of Semgrex search hits?zAIf you want to visualize all search results, leave this unmarked.help)NNz2Which search hit should visualizations start from?zDIf you want to visualize the first 10 search results, set this to 0.r)r, min_valuez/Which search hit should visualizations stop on?zIf you want to visualize the first 10 search results, set this to 11. The 11th result will NOT be displayed. )r,valuer-)rcheckbox number_input) show_window start_window end_windowrrrget_semgrex_window_input;s$ r6cCstjddd}|S)zP Prompts client for whether they want to see xpos tags instead of upos. z&Would you like to visualize xpos tags?zThe default visualization options use upos tags for part-of-speech labeling. If xpos tags aren't available for the sentence, displays upos.r+)rr1)use_xposrrr get_pos_inputWsr8cCs*t\}}t}t}t}|||||fS)zH Tie together all inputs to query user for all possible inputs. )rr*r6r8)rr client_files window_inputvisualize_xposrrr get_input`s r<Trrr9r3clickedpiper4r5r; show_successc  Cs|r|s |s tddS|r|rtddS|s!tddSztddd|dD} |r^gdtd } } } |rI|d |d } } |D]}t|| d | | ||d }| |7} qKn|rpt|| d |d |d ||d } n t|| d td ||d } t| dkrtd| D]}t|}t j |ddddq| rt| d krt dt| dnt dt| dWdWdSWdWdSWdWdS1swYWdSt ytdYdSwdS)aL Run Semgrex search on the input text/files with input query and serve the HTML on the app. @param input_txt: Text to analyze and draw sentences from. @param input_queries: Semgrex queries to parse the input with. @param client_files: Alternative to input text, we can parse the content of files for scaled analysis. @param show_window: Whether or not the user wants a splice of the visualizations @param clicked: Whether or not the button has been clicked to run Semgrex search @param pipe: NLP pipeline to process input with @param start_window: If displaying a splice of visualizations, this is the start idx @param end_window: If displaying a splice of visualizations, this is the end idx @param visualize_xpos: Set to true if using xpos tags for part of speech labels, otherwise use upos tags z9Please provide a text input or upload files for analysis.zQPlease only choose to visualize your input text or your uploaded files, not both.z(Please provide a set of Semgrex queries. Processing...cSsg|]}|qSr)strip).0queryrrr sz'run_semgrex_process..,rinfr/en) start_match end_matchr>r;)rIr>r;No Semgrex match hits!Theightwidth scrollingzCompleted! Visualized z Semgrex search hit.z Semgrex search hits.NzQYour text input or your provided Semgrex queries are incorrect. Please try again.) rerrorspinnersplitfloatrlenwriter componentshtmlsuccessOSError)rrr9r3r=r>r4r5r;r?queries html_strings begin_viz_idx end_viz_idx client_fileclient_file_html_stringss s_no_overflowrrrrun_semgrex_processks      57& > rcc Cshtdd}tj|ddt\}}}}}|\}}}tjddd} t||||| tjd|||d d S) aP Contains the Semgrex portion of the webpage. This contains the markdown and calls to the processes which run when a query is made. When the `Load Semgrex search visualization` button is pressed, the function `run_semgrex_process` is called inside this function and the rendered visual is placed onto the webpage. zDisplaying Semgrex QuerieszE

Enter a text below, along with your Semgrex query of choice.

Tunsafe_allow_htmlz!Load Semgrex search visualizationzsSemgrex search visualizations only display sentences with a query match. Non-matching sentences are not shown.r+pipeline) rrr9r3r=r>r4r5r;N)rtitlerr<buttonrc session_state) html_stringrrr9r:r;r3r4r5r=rrr semgrex_states*   rkcCsPtdtjdttd}tjdddd}tjdddd}tdtjd d d d }g}|D]}t|d }| }| |q.t d}t d} |rzwt dh|} |g} t || | } tj|d} ddd| jD}d}tj|d dtt|| |dgtjdddddd t| dkrtd| D]}d}tj|d dt|}tj|dd d d!qWd"n1swYWntytd#Ynw| r&t|dkrtd$t d%3|d}tj|d} t| ||g}t| |}t|d}d&|}tjd'|d(d)Wd"d"S1swYd"Sd"S)*af Contains the ssurgeon state for the webpage. This contains the markdown and calls the processes that run Ssurgeon operations. When the text boxes, buttons, or other interactable features are edited by the user, this function runs with the updated page state and conducts operations (e.g. runs a Ssurgeon operation on a submitted file) zDisplaying Ssurgeon Resultsr r rz!{}=source >nsubj {} >csubj=bad {}zSsurgeon commandsz&relabelNamedEdge -edge bad -reln advclz***Alternatively, upload file(s) to edit.**TrrrzLoad Ssurgeon visualizationz Edit Filer@) input_str cSsg|] }|jD]}|jqqSr)wordstext)rBsentencewordrrrrD;sz"ssurgeon_state..z'

Previous deprel visualization:

rdFrfr/r.) rrr=r3r9r>r4r5r;r?rrJz%

Edited deprel visualization:

rKrLrMNzZYour text input or your provided Semgrex/Ssurgeon queries are incorrect. Please try again.z$You must provide files for analysis.z Editing... zDownload your edited filezSSurgeon.conll)data file_name) rrgrSAMPLE_SSURGEON_DOCrr rr!r"r#r$rhrRssv,visualize_ssurgeon_deprel_adjusted_str_inputr conll2docjoin sentencesrWrXrcrirUrVrrZrQprocess_doc_one_operationconvert_response_to_doc doc2conlldownload_button)rsemgrex_input_queriesssurgeon_input_queriesr%r&r'r(r)r=clicked_for_file_editsemgrex_queriesssurgeon_queriesr\doc string_txtrjrarb single_filessurgeon_response updated_docoutput output_strrrrssurgeon_states         !      $rcCst}|jdttjdddd|}|j}|tjd<tjddur/t dt ddt j vr?tddd }|t j d<ttdS) Nz --CLASSPATH CLASSPATHzPath to your CoreNLP directory.)typedefaultr,zEProvide a valid $CLASSPATH value (path to your CoreNLP installation).rfrGztokenize, pos, lemma, depparse) processors)argparseArgumentParser add_argumentstrosenvironget parse_argsrloggingrQ ValueErrorrrirrkr)parserargsr en_nlp_stanzarrrmainis(      r__main__)T)*rsys streamlitrstreamlit.components.v1rWv1.stanza.utils.visualization.ssurgeon_visualizerutils visualizationssurgeon_visualizerrwr-stanza.utils.visualization.semgrex_visualizerrr$stanza.utils.visualization.constantsstanza.utils.conllrstanza.server.ssurgeonstanza.pipeline.coreriortypingrr r rrrr*boolintr6r8r<rcrkrr__name__rrrrsf     *    l(j