Initial import

app/models/DocMeta.php

@@ -0,0 +1,13 @@
+<?php
+
+class DocMeta extends Model {
+
+    protected $_classes = [
+        "document" => "Document",
+        "metadata" => "MetaType"
+    ];
+
+
+	
+
+}

app/models/DocProduct.php

@@ -0,0 +1,17 @@
+<?php
+
+class DocProduct extends Model {
+	protected $_classes = [
+		"document" => "Document",
+		"product" => "Product"
+	];
+
+
+	protected $_model = [
+		"id" => [MODEL_SERIAL],
+		"created" => [MODEL_BIGINT, 11],
+		"updated" => [MODEL_BIGINT, 11],
+		"document" => [MODEL_OBJECT, "Document"],
+		"product" => [MODEL_OBJECT, "Product"]
+	];
+}

app/models/Document.php

@@ -0,0 +1,191 @@
+<?php
+
+class Document extends Model {
+
+	protected $_computed = [
+		"related" => "get_related",
+		"revisions" => "get_revisions",
+		"products" => "get_products",
+		"metadata" => "get_metadata",
+		"attachments" => "get_attachments",
+	];
+
+	public function get_revisions() {
+		return Revision::find([["document", "=", $this->id]])->orderBy("revno")->all();
+	}
+
+	public function get_related() {
+		$revs = Revision::find([["document", "=", $this->id]])->all();
+		$related = new Collection;
+		foreach ($revs as $rev) {
+			$q = DB::getInstance()->query("
+				select 
+					id,
+					internal_id,
+					title,
+					subtitle,
+					subsubtitle 
+				from 
+					document 
+				where 
+					internal_id in (
+						select 
+							distinct words.word 
+						from 
+							words,
+							revwords 
+						where 
+							words.id=revwords.word 
+						and 
+							revwords.revision=:rev 
+						and 
+							words.word like '%-%-%'
+			)", array("rev" => $rev->id));
+			while ($r = DB::getInstance()->nextRecord($q)) {
+				if ($r->id != $this->id) {
+					$related->push(new Document($r->id));
+				}
+			}
+		}
+		return $related;
+	}
+
+    public function overview_md() {
+        return \Michelf\Markdown::defaultTransform($this->overview);
+    }
+
+	public function on_delete() {
+		DB::getInstance()->query("delete from docproduct where document=:id", ["id" => $this->id]);
+	}
+
+	private $_products = null;
+
+	public function get_products() {
+		if ($this->_products == null) {
+
+			$this->_products = new Collection;
+			$dpl = DocProduct::find([["document", "=", $this->id]])->all();
+			foreach ($dpl as $dp) {
+				$dp->load("product");
+				$this->_products->push($dp->product);
+			}
+		}
+
+		return $this->_products;
+	}
+
+	public function duplicate() {
+		$newdoc = new Document();
+		$newdoc->title = $this->title;
+		$newdoc->subtitle = $this->subtitle;
+		$newdoc->subsubtitle = $this->subsubtitle;
+		$newdoc->overview = $this->overview;
+		$newdoc->internal_id = $this->internal_id;
+		$newdoc->owner = $this->owner;
+		$newdoc->year = $this->year;
+		$newdoc->month = $this->month;
+
+		$newdoc->save();
+
+		$dpl = DocProduct::find([["document", "=", $this->id]])->all();
+		foreach ($dpl as $dp) {
+			$ndp = new DocProduct();
+			$ndp->document = $newdoc->id;
+			$ndp->product = $dp->product;
+			$ndp->save();
+		}
+		return $newdoc;
+	}
+
+	public function remove_product($id) {
+		$dp = DocProduct::find([["document", "=", $this->id], ["product", "=", $id]])->first();
+		if ($dp) {
+			$dp->delete();
+		}
+	}
+
+	public function set_metadata($metadata, $value) {
+		$m = DocMeta::find([["document", "=", $this->id], ["metadata", "=", $metadata]])->first();
+		if (!$m) {
+			$m = new DocMeta();
+			$m->document = $this->id;
+			$m->metadata = $metadata;
+		}
+		$m->data = $value;
+		$m->save();
+		return $m->id;
+	}
+
+	public function get_metadata() {
+        return DocMeta::find([["document", "=", $this->id]])->all();
+	}
+
+	public function remove_metadata($metadata) {
+		$m = DocMeta::find([["document", "=", $this->id], ["metadata", "=", $metadata]])->first();
+		if ($m) {
+			$m->delete();
+		}
+	}
+
+	public function get_metadata_by_id($metadata) {
+		$m = DocMeta::find([["document", "=", $this->id], ["metadata", "=", $metadata]])->first();
+		if (!$m) {
+			return "";
+		}
+		return $m->data;
+	}
+
+	public function guess_docid() {
+		$text = $this->title . " " . $this->subtitle . " " . $this->subsubtitle;
+		$text .= $this->overview;
+
+		$words = $this->get_words($text);
+
+		return IDMatch::find_docid($words)[0];
+	}
+
+
+	function get_words($text) {
+	    $text = str_replace("\r", " ", $text);
+	    $text = str_replace("\n\n", "\n", $text);
+	    $text = str_replace("\n\n", "\n", $text);
+	    $text = str_replace("\n\n", "\n", $text);
+	    $text = str_replace("\n\n", "\n", $text);
+	    $text = str_replace("\n", " ", $text);
+	    $text = str_replace("\"", "", $text);
+	    $text = str_replace(",", " ", $text);
+	    $text = str_replace("—", "-", $text);
+	    $text = str_replace("~", "-", $text);
+	    $text = str_replace("--", "-", $text);
+	
+	    $text = preg_replace('/([^a-zA-Z\-0-9]+)/', ' ', $text);
+	
+	    //$text = preg_replace("/([A-Za-z])(- )/", '$1', $text);
+	
+	    $words = preg_split('/\s+/', $text);
+	    return $words;
+	}
+
+
+    function get_attachments() {
+        $ap = ROOT . "/attachments/" . $this->id;
+        $files = new Collection;
+        if (!file_exists($ap)) {
+            return $files;
+        }
+        if (!is_dir($ap)) {
+            return $files;
+        }
+
+        $dir = opendir($ap);
+        while ($f = readdir($dir)) {
+            if (substr($f, 0, 1) == ".") continue;
+            $f = new File($ap . "/" . $f);
+            $files->push($f);
+        }
+    
+        return $files;
+    }
+
+
+}

app/models/IDMatch.php

@@ -0,0 +1,54 @@
+<?php
+
+class IDMatch extends Model {
+
+	public static function find_docid($words) {
+		if (!is_array($words)) {
+			$words = IDMatch::get_words($words);
+		}
+		$matches = IDMatch::find()->orderBy("weight")->all();
+
+		foreach ($matches as $match) {
+			$preg = '/^' . $match->regex . '$/';
+
+			foreach ($words as $word) {
+				$word = strtoupper($word);
+				$word = str_replace("(", "C", $word);
+				$word = str_replace("=", "-", $word);
+				$word = str_replace("--", "-", $word);
+
+				if (preg_match($preg, $word, $m)) {
+					return [$m[1], @$m[2]];
+				}
+			}
+		}
+		return false;
+	}
+
+    public static function get_words($text) {
+		print("<<$text>>\n");
+        $text = str_replace("\r", " ", $text);
+        $text = str_replace("\n\n", "\n", $text);
+        $text = str_replace("\n\n", "\n", $text);
+        $text = str_replace("\n\n", "\n", $text);
+        $text = str_replace("\n\n", "\n", $text);
+        $text = str_replace("\n", " ", $text);
+        $text = str_replace("\"", "", $text);
+        $text = str_replace(",", " ", $text);
+        $text = str_replace("–", "-", $text);
+        $text = str_replace("—", "-", $text);
+        $text = str_replace("~", "-", $text);
+        $text = str_replace("--", "-", $text);
+
+        $text = preg_replace('/([^a-zA-Z\-0-9]+)/', ' ', $text);
+
+		print(">>$text<<\n");
+
+        //$text = preg_replace("/([A-Za-z])(- )/", '$1', $text);
+
+        $words = preg_split('/\s+/', $text);
+        return $words;
+    }
+
+
+}

app/models/MetaType.php

@@ -0,0 +1,11 @@
+<?php
+
+class MetaType extends Model {
+	protected $table = "metatypes";
+
+
+	public static function name($id) {
+		$m = new MetaType($id);
+		return $m->name;
+	}
+}

app/models/OCR.php

@@ -0,0 +1,4 @@
+<?php
+
+class OCR extends Model {
+}

app/models/Product.php

@@ -0,0 +1,163 @@
+<?php
+
+class Product extends Model {
+
+	protected $_classes = [
+		"parent" => "Product"
+	];
+
+	protected $_computed = [
+		"documents" => "get_documents",
+		"children" => "get_children",
+	];
+
+	protected $_triggers = [
+		"parent" => "update_path",
+		"title" => "update_path",
+	];
+
+	private $_children = null;
+	public function get_children() {
+		$c = $this->cache_get("children");
+		if ($c) return $c;
+
+		if ($this->_children == null) {
+			$this->_children = Product::find([["parent", "=", $this->id]])->orderBy("title")->all();
+		}
+		$this->cache_set("children", $this->_children);
+		return $this->_children;
+	}
+
+	public function get_full_title() {
+		$tree = $this->get_tree();
+
+		$n = [];
+		foreach ($tree as $t) {
+			$n[] = $t->title;
+		}
+
+		$out = implode(" / ", $n);
+		return $out;
+	}
+
+	public function update_path($ppath = null) {
+		if ($ppath == null) {
+			$this->full_path = $this->get_full_title();
+		} else {
+			$this->full_path = $ppath . " / " . $this->title;
+		}
+		$this->full_path = str_replace("/ / ", "/ ", $this->full_path);
+		$this->save();
+		foreach ($this->get_children() as $child) {
+			$child->update_path($this->fill_path);
+		}
+	}
+
+	public function get_tree() {
+		$out = [];
+		if ($this->load("parent")) {
+			$p = $this->parent;
+			$out = $p->get_tree();
+		}
+		array_push($out, $this);
+
+		return $out;
+	}
+
+	public function overview_md() {
+		return \Michelf\Markdown::defaultTransform($this->overview);
+	}
+
+    public function on_delete() {
+        DB::getInstance()->query("delete from docproduct where product=:id", ["id" => $this->id]);
+    }
+
+	private $_documents = null;
+
+	public function get_documents() {
+
+		$d = $this->cache_get("documents");
+		if ($d) return $d;
+
+		if ($this->_documents == null) {
+			//$dpl = DocProduct::find([["product", "=", $this->id]])->limit(100)->all();
+			$dpl = DocProduct::find([["product", "=", $this->id]])->all();
+			$this->_documents = new Collection;
+			foreach ($dpl as $dp) {
+				if ($dp->load("document")) {
+					$this->_documents->push($dp->document);
+				}
+			}
+
+			$this->_documents->sort("subsubtitle", true);
+			$this->_documents->sort("subtitle", true);
+			$this->_documents->sort("title", true);
+		}
+		$this->cache_set("documents", $this->_documents);
+		return $this->_documents;
+	}
+
+	public function add_document($doc) {
+		$dp = new DocProduct;
+		$dp->product = $this->id;
+		$dp->document = $doc->id;
+		$dp->save();
+	}
+
+	public function meta() {
+		if ($this->metadata == null) {
+			return [];
+		}
+		return explode(",", $this->metadata);
+	}
+
+	public function add_meta($id, $save = true) {
+		$m = $this->meta();
+		if (!in_array($id, $m)) {
+			$m[] = $id;
+		}
+		$this->metadata = implode(",", $m);
+		if ($save) $this->save();
+	}
+
+	public function del_meta($id, $save = true) {
+		$m = $this->meta();
+		$o = [];
+		foreach ($m as $v) {
+			if ($v != $id) {
+				$o[] = $v;
+			}
+		}
+		$this->metadaya = implode(",", $o);
+		if ($save) $this->save();
+	}
+
+	public function documents_sorted_by_meta() {
+
+		if ($this->metadata == null) {
+			$meta = [];
+		} else {
+			$meta = explode(",", $this->metadata);
+		}
+
+		$docs = $this->get_documents();
+
+
+		while (count($meta) > 0) {
+			$mid = array_pop($meta);
+			$docs->sort_with_function( function($a, $b) use ($mid) {
+				$va = $a->get_metadata_by_id($mid);
+				$vb = $b->get_metadata_by_id($mid);
+
+				if ($va > $vb) return 1;
+				if ($va < $vb) return -1;
+				return 0;
+				});
+		}
+
+
+		return $docs;
+
+	}
+
+}

app/models/Revision.php

@@ -0,0 +1,114 @@
+<?php
+
+class Revision extends Model {
+
+	protected $_classes = [
+		"document" => "Document"
+	];
+
+	protected $_computed = [
+		"body" => "get_body",
+	];
+
+	protected $transform = [
+		"info" => "json"
+	];
+
+	function path() {
+		$rs = sprintf("%011d", $this->id);
+		$d1 = substr($rs, 0, 2);
+		$d2 = substr($rs, 2, 3);
+		$d3 = substr($rs, 5, 3);
+		$d4 = substr($rs, 8, 3);
+		$p = sprintf("%s/pdf/%s/%s/%s/%s", ROOT, $d1, $d2, $d3, $d4);
+		return $p;
+	}
+
+	function filename() {
+		if ($this->document) {
+			$out = $this->document->internal_id;
+			if (($this->revno != "") && ($this->revno != "0")) {
+				$out .= "-";
+				$out .= $this->revno;
+			}
+			$out .= " ";
+			$out .= $this->document->title . " " . $this->document->subtitle . " " . $this->document->subsubtitle;
+		} else {
+			$out = "doc";
+		}
+		$out = trim($out);
+		$out.= ".pdf";
+		$out = str_replace(" ", "_", $out);
+		$out = str_replace("/", "_", $out);
+		return $out;
+	}
+
+	function create_cover() {
+		$p = $this->path();
+		$pdf = new PDF($p . "/doc.pdf");
+		$cover = $p . "/cover.jpg";
+		$pdf->extract_page(0, $cover);
+	}
+
+	function cover($size = null) {
+		$f = new File($this->path() . "/cover.jpg");
+		if (!$f->exists()) {
+			$this->create_cover();
+		}
+		if ($size == null) {
+			return "/cover/" . $this->id . "/cover.jpg";
+		}
+		return "/cover/" . $this->id . "/" . $size . "/cover.jpg";
+	}
+
+	private $_body = null;
+
+	public function get_body() {
+		if ($this->_body == null) {
+			$ocr = OCR::find([["revision", "=", $this->id]])->first();
+			if (!$ocr) {
+				return "";
+			}
+			$this->_body = $ocr->body;
+		}
+		return $this->_body;
+	}
+
+	public function get_page($page, $dpi = 300) {
+		$file = new File(sprintf("%s/pages/%04d-%d.jpg", $this->path(), $page, $dpi));
+		if ($file->exists()) {
+			return new Image($file);
+		}
+
+		$dir = sprintf("%s/pages", $this->path());
+		if (!file_exists($dir)) {
+			mkdir($dir, 0777);
+		}
+
+		$pdf = new PDF(sprintf("%s/doc.pdf", $this->path()));
+		$img = $pdf->extract_page($page, $file->path(), $dpi);
+
+		if ($img) {
+			return $img;
+		}
+		return null;
+	}
+
+	public function purge() {
+		$dir = opendir($this->path());
+		while ($file = readdir($dir)) {
+			if (str_ends_with($file, ".jpg")) {
+				unlink($this->path() . "/" . $file);
+			}
+		}
+		closedir($dir);
+
+		$dir = opendir($this->path() . "/pages");
+		while ($file = readdir($dir)) {
+			if (str_ends_with($file, ".jpg")) {
+				unlink($this->path() . "/pages/" . $file);
+			}
+		}
+		closedir($dir);
+	}
+}

app/models/Spider.php

@@ -0,0 +1,4 @@
+<?php
+
+class Spider extends Model {
+}

app/models/SpiderBanned.php

@@ -0,0 +1,4 @@
+<?php
+
+class SpiderBanned extends Model {
+}

app/models/SpiderDom.php

@@ -0,0 +1,4 @@
+<?php
+
+class SpiderDom extends Model {
+}

app/models/SpiderPage.php

@@ -0,0 +1,5 @@
+<?php
+
+class SpiderPage extends Model {
+	protected $table="pages";
+}

app/models/User.php

@@ -0,0 +1,5 @@
+<?php
+
+class User extends Model {
+	public static $table = "users";
+}