Initial import

app/jobs/GeminiJob.php

@@ -0,0 +1,214 @@
+<?php
+
+class GeminiJob extends Job {
+
+	public $docid;
+
+	public function __construct($docid, $source="unknown") {
+		parent::__construct($source);
+		$this->docid = $docid;
+	}
+
+	public function run() {
+
+		$pdf = false;
+
+       $subs = [
+            "Installation and Operating Information",
+            "Installation and Configuration",
+            "Installing and Getting Started",
+            "Installation/Operator's Manual",
+            "Installation/Operator's Guide",
+            "Programmer's Reference Guide",
+			"Field Maintenance Print Set",
+            "Illustrated Parts Breakdown",
+            "Installation/Owner's Guide",
+            "Installation Information",
+            "Installation/User Guide",
+            "User Documentation Kit",
+            "Programmer Information",
+            "Technical Description",
+            "Operator Information",
+            "Configuration Guide",
+            "Upgrade Information",
+            "Installation Manual",
+            "Service Information",
+            "Installation Guide",
+			"Programming Manual",
+            "Maintenance Manual",
+            "Maintenance Guide",
+            "Technical Summary",
+            "Operator's Guide",
+            "System Reference",
+            "User Information",
+            "Technical Manual",
+            "Language Manual",
+            "Service Manual",
+            "Service Guide",
+            "Read Me First",
+            "Owner's Guide",
+            "Release Notes",
+            "Options Guide",
+            "Users' Manual",
+            "User's Manual",
+            "HiTest Notes",
+            "User's Guide",
+            "Design Guide",
+            "User Manual",
+            "User Guide",
+
+        ];
+
+		$doc = new Document($this->docid);
+
+		$minsize = 999999999999999;
+
+		foreach ($doc->revisions as $r) {
+			$p = new PDF($r->path() . "/doc.pdf");
+			if ($p->size() < $minsize) {
+				$minsize = $p->size();
+				$rev = $r;
+				$pdf = $p;
+			}
+		}
+
+		if (!$pdf) {
+			$this->fail();
+			return;
+		}
+
+		$this->status("Processing document with Gemini");
+		try {
+			$gemini = new Gemini();
+			$gemini->upload_callback([$this, "uploadcb"]);
+			$gemini->process_callback([$this, "processcb"]);
+			$lines = $gemini->geminiOverview($pdf);
+		} catch (Exception $e) {
+			$this->status($e->getMessage());
+			if ($e->getmessage() == "The request timed out. Please try again.") {
+				$this->retry();
+			} else if ($e->getmessage() == "HTTP Error 0 requesting AI assistance") {
+				$this->retry();
+			} else {
+				$this->fail();
+			}
+			return;
+		}
+
+		$this->status("Postprocessing returned data");
+		$ld = explode("\n", $lines);
+		if (preg_match('/^\{(.*)\}$/', trim($ld[0]), $m)) {
+			$title = trim($m[1]);
+
+			$title = str_replace("’", "'", $title);
+
+			$newsub = "";
+			$newsubsub = "";
+
+
+			$b = explode(":", $title);
+			if (count($b) > 1) {
+				$title = trim(array_shift($b));
+				$newsub = trim(array_shift($b));
+				$newsubsub = implode(": ", $b);
+			}
+
+			if ($newsub == "") {
+				foreach ($subs as $sub) {
+					if (str_ends_with($title, $sub)) {
+						$newsub = $sub;
+						$title = substr($title, 0, 0 - (strlen($sub) + 1));
+						break;
+					}
+				}
+			}
+		   $doc->title = $title;
+			$doc->subtitle = $newsub;
+			$doc->subsubtitle = $newsubsub;
+
+			array_shift($ld);
+			if (trim($ld[0]) == "") {
+				array_shift($ld);
+			}
+		}
+		if (preg_match('/^\[(.*)\]$/', trim($ld[0]), $m)) {
+
+			$iid = IDMatch::find_docid($this->cleanup($m[1]));
+
+			if ($iid) {
+				$doc->internal_id = $iid[0];
+			}
+			array_shift($ld);
+		}
+		$doc->overview = implode("\n", $ld);
+
+		$doc->save();
+/*
+		$db = DB::getInstance();
+
+		$q = $db->query("
+			select
+				product.id as id,
+				locate(product.title, document.title) as wordoffset
+			from
+				product,
+				document
+			where
+				locate(product.title, document.title) > 0 and
+				document.id = :id and
+				length(product.title) > 3
+			order by
+				length(product.title) desc
+			limit 0,1
+		", ["id" => $doc->id]);
+
+	   if ($r = $db->nextRecord($q)) {
+			$d = new DocProduct();
+			$d->document = $doc->id;
+			$d->product = $r->id;
+			$d->save();
+		} else {
+			$q = $db->query("
+				select
+					product.id as id,
+					locate(product.title, document.overview) as wordoffset
+				from
+					product,
+					document
+				where
+					locate(product.title, document.overview) > 0 and
+					locate(product.title, document.overview) < 300 and
+					document.id = :id and
+					length(product.title) > 3
+				order by
+					length(product.title) desc
+				limit 0,1
+			", ["id" => $doc->id]);
+
+			if ($r = $db->nextRecord($q)) {
+				$d = new DocProduct();
+				$d->document = $doc->id;
+				$d->product = $r->id;
+				$d->save();
+			}
+		}
+*/
+
+		$this->status("Finished");
+	}
+
+
+	public function uploadcb($percent) {
+		$this->status("Uploading: " . $percent . "% complete");
+	}
+
+	public function processcb($message) {
+		$this->status($message);
+	}
+
+	function cleanup($txt) {
+		$txt = str_replace("Ø", "0", $txt);
+		$txt = str_replace(".", " ", $txt);
+		return $txt;
+	}
+}