check authors and remove ghost books

.gitignore

@@ -1 +1,2 @@
 /target
+.vscode

Cargo.lock

@@ -148,7 +148,7 @@ dependencies = [
 
 [[package]]
 name = "pbdbfixer"
-version = "0.2.0"
+version = "0.3.0"
 dependencies = [
  "rusqlite",
  "xml-rs",

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "pbdbfixer"
-version = "0.2.0"
+version = "0.3.0"
 authors = ["Martin Brodbeck <martin@brodbeck-online.de>"]
 edition = "2018"
 

src/main.rs

@@ -1,14 +1,22 @@
 mod pocketbook;
 
 use rusqlite::{named_params, Connection, Result, Transaction, NO_PARAMS};
-use std::error::Error;
-use std::fs::File;
 use std::io::BufReader;
+use std::{collections::HashMap, fs::File};
+use std::{error::Error, io::Read};
 use xml::reader::{EventReader, ParserConfig, XmlEvent};
 use zip::{read::ZipFile, ZipArchive};
 
-fn get_root_file(container: ZipFile) -> Result<Option<String>, Box<dyn Error>> {
+fn get_root_file(mut container: ZipFile) -> Result<Option<String>, Box<dyn Error>> {
-    let parser = EventReader::new(container);
+    let mut buf = String::new();
+    container.read_to_string(&mut buf).unwrap();
+
+    // Get rid of the BOM mark, if any
+    if buf.starts_with("\u{feff}") {
+        buf = buf.strip_prefix("\u{feff}").unwrap().to_owned();
+    }
+
+    let parser = EventReader::new(BufReader::new(buf.as_bytes()));
 
     for e in parser {
         match e {
@@ -30,6 +38,11 @@ fn get_root_file(container: ZipFile) -> Result<Option<String>, Box<dyn Error>> {
     Ok(None)
 }
 
+struct Refine {
+    role: String,
+    file_as: String,
+}
+
 fn get_attribute_file_as(opf: ZipFile) -> Option<String> {
     let parser = ParserConfig::new()
         .trim_whitespace(true)
@@ -37,10 +50,12 @@ fn get_attribute_file_as(opf: ZipFile) -> Option<String> {
         .coalesce_characters(true)
         .create_reader(opf);
 
-    let mut refines_found = false;
-    let mut refines_entries = Vec::new();
     let mut is_epub3 = false;
     let mut creator_ids = Vec::new();
+    let mut refines_found = false;
+    let mut role_found = false;
+    let mut refine_entries = HashMap::new();
+    let mut curr_id = String::new();
 
     for e in parser {
         match e {
@@ -77,11 +92,43 @@ fn get_attribute_file_as(opf: ZipFile) -> Option<String> {
                     .any(|attr| attr.name.local_name == "property" && attr.value == "file-as")
                 {
                     refines_found = true;
+                    curr_id = attributes
+                        .iter()
+                        .find(|a| a.name.local_name == "refines")
+                        .unwrap()
+                        .value
+                        .clone();
+                } else if attributes.iter().any(|attr| {
+                    attr.name.local_name == "refines" && creator_ids.contains(&attr.value)
+                }) && attributes
+                    .iter()
+                    .any(|attr| attr.name.local_name == "property" && attr.value == "role")
+                {
+                    role_found = true;
+                    curr_id = attributes
+                        .iter()
+                        .find(|a| a.name.local_name == "refines")
+                        .unwrap()
+                        .value
+                        .clone();
                 }
             }
             Ok(XmlEvent::Characters(value)) => {
-                if refines_found == true {
+                if role_found == true {
-                    refines_entries.push(value);
+                    if value == "aut" {
+                        let entry = refine_entries.entry(curr_id.clone()).or_insert(Refine {
+                            role: "".to_string(),
+                            file_as: "".to_string(),
+                        });
+                        entry.role = value;
+                    }
+                    role_found = false;
+                } else if refines_found == true {
+                    let entry = refine_entries.entry(curr_id.clone()).or_insert(Refine {
+                        role: "".to_string(),
+                        file_as: "".to_string(),
+                    });
+                    entry.file_as = value;
                     refines_found = false;
                 }
             }
@@ -97,10 +144,156 @@ fn get_attribute_file_as(opf: ZipFile) -> Option<String> {
         }
     }
 
-    if refines_entries.len() == 1 {
+    if refine_entries.len() == 1 {
-        return Some(refines_entries.remove(0));
+        return Some(refine_entries.values().next().unwrap().file_as.clone());
-    } else if refines_entries.len() >= 2 {
+    } else if refine_entries.len() >= 2 {
-        return Some(refines_entries.join(" & "));
+        return Some(
+            refine_entries
+                .values()
+                .into_iter()
+                .filter(|v| v.role == "aut")
+                .map(|v| v.file_as.clone())
+                .collect::<Vec<String>>()
+                .join(" & "),
+        );
+    }
+
+    None
+}
+
+struct Creator {
+    role: String,
+    name: String,
+}
+
+fn get_attribute_creator(opf: ZipFile) -> Option<String> {
+    let parser = ParserConfig::new()
+        .trim_whitespace(true)
+        .ignore_comments(true)
+        .coalesce_characters(true)
+        .create_reader(opf);
+
+    let mut is_epub3 = false;
+    let mut creator_found = true;
+    let mut creator_ids = Vec::new();
+    let mut role_found = false;
+    let mut creator_entries = HashMap::new();
+    let mut epub2_creator_entries = Vec::new();
+    let mut curr_id = String::new();
+
+    for e in parser {
+        match e {
+            Ok(XmlEvent::StartElement {
+                name, attributes, ..
+            }) if name.local_name == "package" => {
+                for attr in attributes {
+                    if attr.name.local_name == "version" {
+                        if attr.value.starts_with("3") == true {
+                            is_epub3 = true;
+                        }
+                    }
+                }
+            }
+            Ok(XmlEvent::StartElement {
+                name, attributes, ..
+            }) if name.local_name == "creator" => {
+                creator_found = true;
+                if !is_epub3 {
+                    match attributes
+                        .iter()
+                        .find(|attr| attr.name.local_name == "role")
+                    {
+                        Some(attr) => {
+                            epub2_creator_entries.push(Creator {
+                                role: attr.value.clone(),
+                                name: "".to_string(),
+                            });
+                        }
+                        None => {
+                            epub2_creator_entries.push(Creator {
+                                role: "aut".to_string(),
+                                name: "".to_string(),
+                            });
+                        }
+                    }
+                }
+                for attr in attributes {
+                    if is_epub3 && attr.name.local_name == "id" {
+                        creator_ids.push("#".to_owned() + attr.value.as_str());
+                        //creator_entries.insert(attr.value.clone(), Creator{role: "".to_string(), name: "".to_string()});
+                        curr_id = "#".to_owned() + attr.value.as_str();
+                    }
+                }
+            }
+            Ok(XmlEvent::StartElement {
+                name, attributes, ..
+            }) if name.local_name == "meta" => {
+                if attributes.iter().any(|attr| {
+                    attr.name.local_name == "refines" && creator_ids.contains(&attr.value)
+                }) && attributes
+                    .iter()
+                    .any(|attr| attr.name.local_name == "property" && attr.value == "role")
+                {
+                    role_found = true;
+                    curr_id = attributes
+                        .iter()
+                        .find(|a| a.name.local_name == "refines")
+                        .unwrap()
+                        .value
+                        .clone();
+                }
+            }
+            Ok(XmlEvent::Characters(value)) => {
+                if creator_found && is_epub3 == false {
+                    epub2_creator_entries.last_mut().unwrap().name = value.clone();
+                } else if creator_found && is_epub3 == true {
+                    let entry = creator_entries.entry(curr_id.clone()).or_insert(Creator {
+                        role: "".to_string(),
+                        name: "".to_string(),
+                    });
+                    entry.name = value;
+                    creator_found = false;
+                } else if role_found == true {
+                    let entry = creator_entries.entry(curr_id.clone()).or_insert(Creator {
+                        role: "".to_string(),
+                        name: "".to_string(),
+                    });
+                    entry.role = value;
+                    role_found = false;
+                }
+            }
+            Ok(XmlEvent::StartElement { .. }) => {
+                if creator_found == true {
+                    creator_found = false;
+                }
+            }
+            Err(e) => {
+                println!("{}", e);
+                break;
+            }
+            _ => {}
+        }
+    }
+
+    if !is_epub3 && epub2_creator_entries.len() >= 1 {
+        return Some(
+            epub2_creator_entries
+                .into_iter()
+                .filter(|v| v.role == "aut")
+                .map(|v| v.name.clone())
+                .collect::<Vec<String>>()
+                .join(", "),
+        );
+    } else if creator_entries.len() >= 1 {
+        return Some(
+            creator_entries
+                .values()
+                .into_iter()
+                .filter(|v| v.role == "aut")
+                .map(|v| v.name.clone())
+                .collect::<Vec<String>>()
+                .join(", "),
+        );
     }
 
     None
@@ -109,43 +302,107 @@ fn get_attribute_file_as(opf: ZipFile) -> Option<String> {
 struct BookEntry {
     id: i32,
     filepath: String,
+    author: String,
+    firstauthor: String,
 }
 
-fn fix_firstauthor(tx: &Transaction) -> i32 {
+fn get_epubs_from_database(tx: &Transaction) -> Vec<BookEntry> {
-    let mut authors_fixed = 0;
+    let mut book_entries = Vec::new();
 
-    // Get book ids from entries where we have something like "firstname lastname" in author
+    let mut stmt = tx
-    // but no "lastname, firstname" in fistauthor
+        .prepare(
-    // Get also book ids from the special case where we have multiple authors (separated by ", " in authors)
+            r"
-    // but no ampersand ("&") in firstauthor
+    SELECT books.id, folders.name, files.filename, books.firstauthor, books.author
-    let mut stmt = tx.prepare(r"
+      FROM books_impl books JOIN files
-        SELECT files.book_id, folders.name, files.filename 
+        ON books.id = files.book_id
-          FROM files INNER JOIN folders
+        JOIN folders
-            ON files.folder_id = folders.id
+          ON folders.id = files.folder_id
-          WHERE files.book_id IN 
+      WHERE files.storageid = 1 AND books.ext = 'epub'
-            (
+      ORDER BY books.id",
-              SELECT DISTINCT id FROM books_impl 
+        )
-                WHERE (ext LIKE 'epub' AND author LIKE '% %' AND (firstauthor NOT LIKE '%\,%' ESCAPE '\' OR firstauthor LIKE '%&amp;%'))
+        .unwrap();
-                  OR (ext LIKE 'epub' AND author LIKE '%\, %' ESCAPE '\' AND firstauthor NOT LIKE '%&%')
-            )
-            AND files.storageid = 1
-        ;").unwrap();
 
     let mut rows = stmt.query(NO_PARAMS).unwrap();
-    let mut bookentries = Vec::new();
 
     while let Some(row) = rows.next().unwrap() {
         let book_id: i32 = row.get(0).unwrap();
         let prefix: String = row.get(1).unwrap();
         let filename: String = row.get(2).unwrap();
         let filepath = format!("{}/{}", prefix, filename);
-        bookentries.push(BookEntry {
+        let firstauthor: String = row.get(3).unwrap();
+        let author: String = row.get(4).unwrap();
+
+        let entry = BookEntry {
             id: book_id,
             filepath,
-        });
+            firstauthor,
+            author,
+        };
+
+        book_entries.push(entry);
     }
 
-    for entry in bookentries {
+    book_entries
+}
+
+fn remove_ghost_books_from_db(tx: &Transaction) -> usize {
+    let mut stmt = tx
+        .prepare(
+            r"
+            DELETE FROM books_impl
+            WHERE id IN (
+              SELECT books.id
+                FROM books_impl books
+                  LEFT OUTER JOIN files
+                    ON books.id = files.book_id
+                WHERE files.filename is NULL
+            )",
+        )
+        .unwrap();
+
+    let num = stmt.execute(NO_PARAMS).unwrap();
+
+    tx.execute(
+        r"DELETE FROM books_settings WHERE bookid NOT IN ( SELECT id FROM books_impl )",
+        NO_PARAMS,
+    )
+    .unwrap();
+    tx.execute(
+        r"DELETE FROM books_uids WHERE book_id NOT IN ( SELECT id FROM books_impl )",
+        NO_PARAMS,
+    )
+    .unwrap();
+    tx.execute(
+        r"DELETE FROM bookshelfs_books WHERE bookid NOT IN ( SELECT id FROM books_impl )",
+        NO_PARAMS,
+    )
+    .unwrap();
+    tx.execute(
+        r"DELETE FROM booktogenre WHERE bookid NOT IN ( SELECT id FROM books_impl )",
+        NO_PARAMS,
+    )
+    .unwrap();
+    tx.execute(
+        r"DELETE FROM social WHERE bookid NOT IN ( SELECT id FROM books_impl )",
+        NO_PARAMS,
+    )
+    .unwrap();
+
+    num
+}
+
+struct Statistics {
+    authors_fixed: i32,
+    ghost_books_cleaned: usize,
+}
+
+fn fix_db_entries(tx: &Transaction, book_entries: &Vec<BookEntry>) -> Statistics {
+    let mut stat = Statistics {
+        authors_fixed: 0,
+        ghost_books_cleaned: 0,
+    };
+
+    for entry in book_entries {
         let file = File::open(entry.filepath.as_str());
         let file = match file {
             Err(_) => continue,
@@ -158,29 +415,53 @@ fn fix_firstauthor(tx: &Transaction) -> i32 {
 
         if let Some(opf_file) = get_root_file(container).unwrap() {
             let opf = archive.by_name(opf_file.as_str()).unwrap();
+            // firstauthor…
             if let Some(file_as) = get_attribute_file_as(opf) {
-                let mut stmt = tx
+                if !file_as.split(" & ").all(|s| entry.firstauthor.contains(s)) {
-                    .prepare("UPDATE books_impl SET firstauthor = :file_as WHERE id = :book_id")
+                    let mut stmt = tx
-                    .unwrap();
+                        .prepare("UPDATE books_impl SET firstauthor = :file_as WHERE id = :book_id")
-                stmt.execute_named(named_params![":file_as": file_as, ":book_id": entry.id])
+                        .unwrap();
-                    .unwrap();
+                    stmt.execute_named(named_params![":file_as": file_as, ":book_id": entry.id])
-                authors_fixed = authors_fixed + 1;
+                        .unwrap();
+                    stat.authors_fixed = stat.authors_fixed + 1;
+                }
+            }
+            let opf = archive.by_name(opf_file.as_str()).unwrap();
+            // author…
+            if let Some(creator) = get_attribute_creator(opf) {
+                if !creator.split(", ").all(|s| entry.author.contains(s))
+                    || creator.len() < entry.author.len()
+                {
+                    let mut stmt = tx
+                        .prepare("UPDATE books_impl SET author = :creator WHERE id = :book_id")
+                        .unwrap();
+                    stmt.execute_named(named_params![":creator": creator, ":book_id": entry.id])
+                        .unwrap();
+                    stat.authors_fixed = stat.authors_fixed + 1;
+                }
             }
         }
     }
 
-    authors_fixed
+    // ghost books
+    let num = remove_ghost_books_from_db(tx);
+    stat.ghost_books_cleaned = num;
+
+    stat
 }
 
 fn main() {
     let mut conn = Connection::open("/mnt/ext1/system/explorer-3/explorer-3.db").unwrap();
 
+    conn.execute("PRAGMA foreign_keys = 0", NO_PARAMS).unwrap();
+
     let tx = conn.transaction().unwrap();
-    let authors_fixed = fix_firstauthor(&tx);
+    let book_entries = get_epubs_from_database(&tx);
+    let stat = fix_db_entries(&tx, &book_entries);
     tx.commit().unwrap();
 
     if cfg!(target_arch = "arm") {
-        if authors_fixed == 0 {
+        if stat.authors_fixed == 0 {
             pocketbook::dialog(
                 pocketbook::Icon::Info,
                 "The database seems to be ok.\nNothing had to be fixed.",
@@ -188,8 +469,16 @@ fn main() {
         } else {
             pocketbook::dialog(
                 pocketbook::Icon::Info,
-                &format!("Authors fixed: {}", &authors_fixed),
+                &format!(
+                    "Authors fixed: {}\nBooks cleaned from DB: {}",
+                    &stat.authors_fixed, &stat.ghost_books_cleaned
+                ),
             );
         }
+    } else {
+        println!(
+            "Authors fixed: {}\nBooks cleaned from DB: {}",
+            &stat.authors_fixed, &stat.ghost_books_cleaned
+        );
     }
 }