contractor/process_v2.cfm

<!--- process_v2.cfm — v2 pipeline: LLM-интерпретация ДС → apply_events --->
<cftry>
    <cfquery name="supps" datasource="baza">
        SELECT s.id as supp_id, s.type, d.id as doc_id, d.filename, d.elements_json
        FROM supplements s
        JOIN documents d ON s.document_id = d.id
        WHERE s.contract_id = <cfqueryparam value="#url.contract_id#" cfsqltype="cf_sql_varchar">
            AND d.elements_json IS NOT NULL
        ORDER BY s.created_at
    </cfquery>

    <cfif supps.recordCount EQ 0>
        <cfoutput>data: #serializeJSON({type:"error",message:"Нет распарсенных файлов"})#

</cfoutput><cfflush>
        <cfabort>
    </cfif>

    <cfset totalUnresolved = 0>
    <cfset llmStart = getTickCount()>

    <cfloop query="supps">
        <cfset sid = supps.supp_id>

        <!--- Текущая спецификация из spec_current --->
        <cfquery name="curSpec" datasource="baza">
            SELECT name_hash, name, price, qty, sum, date_start
            FROM spec_current
            WHERE contract_id = <cfqueryparam value="#url.contract_id#" cfsqltype="cf_sql_varchar">
            ORDER BY name
        </cfquery>
        <cfset currentSpec = []>
        <cfloop query="curSpec">
            <cfset arrayAppend(currentSpec, {
                hash: curSpec.name_hash,
                name: curSpec.name,
                price: curSpec.price,
                qty: curSpec.qty,
                sum: curSpec.sum,
                date_start: curSpec.date_start
            })>
        </cfloop>

        <!--- textify elements_json → doc_text --->
        <cfset elements = deserializeJSON(supps.elements_json)>
        <cfset textLines = []>
        <cfloop array="#elements#" index="el">
            <cfif el.type EQ "paragraph">
                <cfset prefix = len(el.style) ? "[#el.style#] " : "">
                <cfset arrayAppend(textLines, prefix & el.text)>
            <cfelseif el.type EQ "table">
                <cfset rows = el.rows>
                <cfif arrayLen(rows) GT 0>
                    <cfset ncols = arrayLen(rows[1])>
                    <cfset arrayAppend(textLines, "--- Таблица (" & arrayLen(rows) & "×" & ncols & ") ---")>
                    <cfloop array="#rows#" index="row">
                        <cfset cells = []>
                        <cfloop from="1" to="#ncols#" index="ci">
                            <cfset val = "">
                            <cfif ci LTE arrayLen(row)>
                                <cfset val = replace(replace(toString(row[ci]), chr(10), " ", "ALL"), "|", "\|", "ALL")>
                            </cfif>
                            <cfset arrayAppend(cells, val)>
                        </cfloop>
                        <cfset arrayAppend(textLines, "| " & arrayToList(cells, " | ") & " |")>
                    </cfloop>
                    <cfset arrayAppend(textLines, "")>
                </cfif>
            </cfif>
        </cfloop>
        <cfset docText = arrayToList(textLines, chr(10))>

        <!--- SSE: начинаем LLM --->
        <cfoutput>data: #serializeJSON({type:"extract_start",supplement_id:sid,filename:supps.filename})#

</cfoutput><cfflush>

        <!--- Вызов ВМ /llm-ops --->
        <cfset t0 = getTickCount()>
        <cfset llmPayload = {
            contract_id: url.contract_id,
            supplement_id: sid,
            current_spec: currentSpec,
            doc_text: docText
        }>

        <cfhttp url="https://contracts.kube5s.ru/llm-ops" method="POST" timeout="130">
            <cfhttpparam type="header" name="Content-Type" value="application/json">
            <cfhttpparam type="body" value="#serializeJSON(llmPayload)#">
        </cfhttp>

        <cfset elapsed = int((getTickCount() - t0) / 1000 * 10 + 0.5) / 10>

        <cfif NOT (cfhttp.statusCode EQ 200 OR left(cfhttp.statusCode, 3) EQ "200")>
            <cfoutput>data: #serializeJSON({type:"extract_error",supplement_id:sid,filename:supps.filename,error:"LLM HTTP #cfhttp.statusCode#",time_s:elapsed})#

</cfoutput><cfflush>
            <cfcontinue>
        </cfif>

        <cftry>
            <cfset llmResult = deserializeJSON(cfhttp.fileContent)>
            <cfset ops = structKeyExists(llmResult, "ops") ? llmResult.ops : []>
            <cfset mode = structKeyExists(llmResult, "mode") ? llmResult.mode : "partial">

            <cfoutput>data: #serializeJSON({type:"llm_done",supplement_id:sid,filename:supps.filename,ops_count:arrayLen(ops),mode:mode,time_s:elapsed})#

</cfoutput><cfflush>

            <!--- Применить операции --->
            <cfset applyPayload = {
                supplement_id: sid,
                ops: ops
            }>

            <cfhttp url="https://contractor.luceek8s.dev.nubes.ru/apply_events.cfm?contract_id=#url.contract_id#&mode=#mode#" method="POST" timeout="30">
                <cfhttpparam type="header" name="Content-Type" value="application/json">
                <cfhttpparam type="body" value="#serializeJSON(applyPayload)#">
            </cfhttp>

            <cfif cfhttp.statusCode EQ 200 OR left(cfhttp.statusCode, 3) EQ "200">
                <cfset applyResult = deserializeJSON(cfhttp.fileContent)>
                <cfif structKeyExists(applyResult, "summary")>
                    <cfset totalUnresolved += applyResult.summary.unresolved>
                    <cfoutput>data: #serializeJSON({type:"applied",supplement_id:sid,summary:applyResult.summary})#

</cfoutput><cfflush>
                <cfelse>
                    <cfoutput>data: #serializeJSON({type:"apply_error",supplement_id:sid,error:structKeyExists(applyResult,'error')?applyResult.error:'unknown'})#

</cfoutput><cfflush>
                </cfif>
            <cfelse>
                <cfoutput>data: #serializeJSON({type:"apply_error",supplement_id:sid,error:"apply HTTP #cfhttp.statusCode#"})#

</cfoutput><cfflush>
            </cfif>

            <cfcatch>
                <cfoutput>data: #serializeJSON({type:"extract_error",supplement_id:sid,filename:supps.filename,error:"JSON parse: " & cfcatch.message,time_s:elapsed})#

</cfoutput><cfflush>
            </cfcatch>
        </cftry>
    </cfloop>

    <!--- Финальное событие --->
    <cfset totalTime = int((getTickCount() - llmStart) / 1000 * 10 + 0.5) / 10>
    <cfoutput>data: #serializeJSON({type:"done",total_time_s:totalTime,total_unresolved:totalUnresolved})#

</cfoutput><cfflush>

    <cfcatch>
        <cfoutput>data: #serializeJSON({type:"error",message:cfcatch.message,detail:cfcatch.detail})#

</cfoutput><cfflush>
    </cfcatch>
</cftry>