chore: bump version

Cargo.toml

@@ -1,19 +1,21 @@
 [package]
 name = "bzkf-rwdp-check"
-version = "0.1.0"
+version = "0.3.2"
 edition = "2021"
-
-# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
+authors = ["Paul-Christian Volkmer <volkmer_p@ukw.de>"]
+description = "Anwendung zur Durchführung einer Plausibilitätsprüfung anhand der Daten für die BZKF Real World Data Platform."
+repository = "https://github.com/CCC-MF/bzkf-rwdp-check"
 
 [dependencies]
-clap = { version = "4.4", features = ["std", "help", "usage", "derive", "error-context"], default-features = false }
+clap = { version = "4.5", features = ["std", "help", "usage", "derive", "error-context"], default-features = false }
 console = "0.15"
 csv = "1.3"
 dialoguer = "0.11"
-itertools = "0.12"
-mysql = "24.0"
+itertools = "0.13"
+mysql = "25.0"
 serde = { version = "1.0", features = ["derive"] }
 urlencoding = "2.1"
+regex = "1.10"
 
 [profile.release]
 opt-level = "s"

README.md

@@ -22,6 +22,8 @@ flowchart LR
 Die Anwendung gibt für die möglichen Quellen der Kennzahlen die Anzahl der _Conditions_, gruppiert nach ICD-10 Gruppen,
 aus.
 
+Unterstützt wird eien OPAL-CSV-Datei (wie für BZKF vorgesehen) und eine Onkostar-Datenbank, basierend auf MariaDB oder MySQL.
+
 ![Ausgabe](docs/screenshot.png)
 
 ## Kennzahlen aus der CSV-Datei
@@ -58,6 +60,13 @@ Options:
   -y, --year <YEAR>          Jahr der Diagnose
 ```
 
+Der zusätzliche Parameter `--ignore-exports-since` ist optional.
+Wird er angegeben, werden keine Einträge mit Exportdatum ab diesem Datum verwendet.
+Dies eignet sich um nachträglich Zahlen zu einem bestimmten Datum zu ermitteln.
+
+Der Parameter `--include-extern` schließt Meldungen mit externer Diagnosestellung ein.
+Diese sind normalerweise nicht enthalten. 
+
 ## Export aus der Onkostar-Datenbank
 
 Die Anwendung ist in der Lage, mit dem Befehl `export` die Spalten
@@ -76,6 +85,7 @@ zusätzlich gibt es noch die folgenden Parameter:
 Options:
       --pat-id               Export mit Klartext-Patienten-ID
   -o, --output <OUTPUT>      Ausgabedatei
+      --xls-csv              Export mit Trennzeichen ';' für Excel
 ```
 
 ## Vergleich CSV-Datei für OPAL und Onkostar-Datenbank

src/cli.rs

@@ -19,6 +19,7 @@
  */
 
 use clap::{Parser, Subcommand};
+use regex::Regex;
 
 #[derive(Parser)]
 #[command(author, version, about)]
@@ -58,6 +59,10 @@ pub enum SubCommand {
         user: String,
         #[arg(short = 'y', long, help = "Jahr der Diagnose")]
         year: String,
+        #[arg(long, value_parser = value_is_date, help = "Ignoriere LKR-Exporte seit Datum")]
+        ignore_exports_since: Option<String>,
+        #[arg(long, help = "Meldungen mit externer Diagnose einschließen")]
+        include_extern: bool,
     },
     #[command(
         about = "Erstellt eine (reduzierte) CSV-Datei zum direkten Vergleich mit der OPAL-CSV-Datei"
@@ -88,6 +93,12 @@ pub enum SubCommand {
         output: String,
         #[arg(short = 'y', long, help = "Jahr der Diagnose")]
         year: String,
+        #[arg(long, value_parser = value_is_date, help = "Ignoriere LKR-Exporte seit Datum")]
+        ignore_exports_since: Option<String>,
+        #[arg(long, help = "Export mit Trennzeichen ';' für Excel")]
+        xls_csv: bool,
+        #[arg(long, help = "Meldungen mit externer Diagnose einschließen")]
+        include_extern: bool,
     },
     #[command(about = "Abgleich zwischen CSV-Datei für OPAL und Onkostar-Datenbank")]
     Compare {
@@ -116,5 +127,21 @@ pub enum SubCommand {
         file: String,
         #[arg(short = 'y', long, help = "Jahr der Diagnose")]
         year: String,
+        #[arg(long, value_parser = value_is_date, help = "Ignoriere LKR-Exporte seit Datum")]
+        ignore_exports_since: Option<String>,
+        #[arg(long, help = "Meldungen mit externer Diagnose einschließen")]
+        include_extern: bool,
     },
 }
+
+fn value_is_date(value: &str) -> Result<String, String> {
+    let re = Regex::new(r"^[0-9]{4}-[0-1][0-9]-[0-3][0-9]$").unwrap();
+    if re.is_match(value) {
+        Ok(value.into())
+    } else {
+        Err(format!(
+            "Ungültiges Datum '{}', bitte im Format 'yyyy-mm-dd' angeben",
+            value
+        ))
+    }
+}

src/common.rs

@@ -68,7 +68,7 @@ impl Check {
             .map(|(icd10, group)| (icd10, group.collect::<Vec<_>>()))
             .map(|record| Icd10GroupSize {
                 name: record.0,
-                size: record.1.iter().count(),
+                size: record.1.len(),
             })
             .collect::<Vec<_>>();
 
@@ -76,10 +76,7 @@ impl Check {
     }
 
     pub fn is_relevant(code: &str) -> bool {
-        match Self::map_icd_code(code).as_str() {
-            "Other" => false,
-            _ => true,
-        }
+        !matches!(Self::map_icd_code(code).as_str(), "Other")
     }
 
     fn map_icd_code(code: &str) -> String {

src/database.rs

@@ -18,6 +18,8 @@
  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
  */
 
+use std::time::Duration;
+
 use mysql::prelude::Queryable;
 use mysql::{params, Pool};
 
@@ -33,13 +35,18 @@ impl DatabaseSource {
         DatabaseSource(url)
     }
 
-    pub fn check(&self, year: &str) -> Result<Vec<Icd10GroupSize>, ()> {
+    pub fn check(
+        &self,
+        year: &str,
+        ignore_exports_since: &str,
+        include_extern: bool,
+    ) -> Result<Vec<Icd10GroupSize>, ()> {
         match Pool::new(self.0.as_str()) {
             Ok(pool) => {
-                if let Ok(mut connection) = pool.get_conn() {
+                if let Ok(mut connection) = pool.try_get_conn(Duration::from_secs(3)) {
                     return match connection.exec_map(
                         SQL_QUERY,
-                        params! {"year" => year},
+                        params! {"year" => year, "ignore_exports_since" => ignore_exports_since, "include_extern" => if include_extern { 1 } else { 0 } },
                         |(icd10_group, count)| Icd10GroupSize {
                             name: icd10_group,
                             size: count,
@@ -58,13 +65,19 @@ impl DatabaseSource {
         Err(())
     }
 
-    pub fn export(&self, year: &str, use_pat_id: bool) -> Result<Vec<ExportData>, ()> {
+    pub fn export(
+        &self,
+        year: &str,
+        ignore_exports_since: &str,
+        use_pat_id: bool,
+        include_extern: bool,
+    ) -> Result<Vec<ExportData>, ()> {
         match Pool::new(self.0.as_str()) {
             Ok(pool) => {
-                if let Ok(mut connection) = pool.get_conn() {
+                if let Ok(mut connection) = pool.try_get_conn(Duration::from_secs(3)) {
                     return match connection.exec_map(
                         EXPORT_QUERY,
-                        params! {"year" => year},
+                        params! {"year" => year, "ignore_exports_since" => ignore_exports_since, "include_extern" => if include_extern { 1 } else { 0 } },
                         |(condition_id, icd_10_code, diagnosis_date, pat_id)| ExportData {
                             condition_id,
                             icd_10_code,

src/main.rs

@@ -23,7 +23,7 @@ use std::path::Path;
 
 use clap::Parser;
 use console::{style, Term};
-use csv::Writer;
+use csv::WriterBuilder;
 use itertools::Itertools;
 
 use crate::cli::{Cli, SubCommand};
@@ -36,6 +36,27 @@ mod database;
 mod opal;
 mod resources;
 
+fn request_password_if_none(password: Option<String>) -> String {
+    if let Some(password) = password {
+        password
+    } else {
+        let password = dialoguer::Password::new()
+            .with_prompt("Password")
+            .interact()
+            .unwrap_or_default();
+        let _ = Term::stdout().clear_last_lines(1);
+        password
+    }
+}
+
+fn sanitize_year(year: String) -> String {
+    if year.len() == 4 {
+        year
+    } else {
+        format!("2{:0>3}", year)
+    }
+}
+
 fn print_items(items: &[Icd10GroupSize]) {
     let term = Term::stdout();
     let _ = term.write_line(
@@ -46,6 +67,38 @@ fn print_items(items: &[Icd10GroupSize]) {
     items.iter().for_each(|item| {
         let _ = term.write_line(&format!("{:<20}={:>6}", item.name, item.size));
     });
+    let sum: usize = items
+        .iter()
+        .filter(|item| item.name != "Other")
+        .map(|item| item.size)
+        .sum();
+    let _ = term.write_line(&style("─".repeat(27)).dim().to_string());
+    let _ = term.write_line(
+        &style(format!("{:<20}={:>6}", "Summe (C**.*/D**.*)", sum))
+            .dim()
+            .to_string(),
+    );
+    let sum: usize = items.iter().map(|item| item.size).sum();
+    let _ = term.write_line(
+        &style(format!("{:<20}={:>6}", "Gesamtsumme", sum))
+            .dim()
+            .to_string(),
+    );
+    let _ = term.write_line(&style("─".repeat(27)).dim().to_string());
+}
+
+fn print_extern_notice(include_extern: bool) {
+    let _ = Term::stdout().write_line(
+        format!(
+            "{} Die Datenbankanfrage schließt Meldungen mit externer Diagnose {}.",
+            style("Hinweis:").bold().underlined(),
+            match include_extern {
+                true => style("ein").yellow(),
+                false => style("nicht ein (Standard)").green(),
+            }
+        )
+        .as_str(),
+    );
 }
 
 fn main() -> Result<(), Box<dyn Error>> {
@@ -65,23 +118,11 @@ fn main() -> Result<(), Box<dyn Error>> {
             port,
             user,
             year,
+            ignore_exports_since,
+            include_extern,
         } => {
-            let password = if let Some(password) = password {
-                password
-            } else {
-                let password = dialoguer::Password::new()
-                    .with_prompt("Password")
-                    .interact()
-                    .unwrap_or_default();
-                let _ = term.clear_last_lines(1);
-                password
-            };
-
-            let year = if year.len() == 4 {
-                year
-            } else {
-                format!("2{:0>3}", year)
-            };
+            let password = request_password_if_none(password);
+            let year = sanitize_year(year);
 
             let _ = term.write_line(
                 &style(format!("Warte auf Daten für das Diagnosejahr {}...", year))
@@ -91,11 +132,16 @@ fn main() -> Result<(), Box<dyn Error>> {
 
             let db = DatabaseSource::new(&database, &host, &password, port, &user);
             let items = db
-                .check(&year)
+                .check(
+                    &year,
+                    &ignore_exports_since.unwrap_or("9999-12-31".into()),
+                    include_extern,
+                )
                 .map_err(|_e| "Fehler bei Zugriff auf die Datenbank")?;
 
             let _ = term.clear_last_lines(1);
 
+            print_extern_notice(include_extern);
             print_items(&items);
         }
         SubCommand::Export {
@@ -107,23 +153,12 @@ fn main() -> Result<(), Box<dyn Error>> {
             user,
             output,
             year,
+            ignore_exports_since,
+            xls_csv,
+            include_extern,
         } => {
-            let password = if let Some(password) = password {
-                password
-            } else {
-                let password = dialoguer::Password::new()
-                    .with_prompt("Password")
-                    .interact()
-                    .unwrap_or_default();
-                let _ = term.clear_last_lines(1);
-                password
-            };
-
-            let year = if year.len() == 4 {
-                year
-            } else {
-                format!("2{:0>3}", year)
-            };
+            let password = request_password_if_none(password);
+            let year = sanitize_year(year);
 
             let _ = term.write_line(
                 &style(format!("Warte auf Daten für das Diagnosejahr {}...", year))
@@ -133,12 +168,24 @@ fn main() -> Result<(), Box<dyn Error>> {
 
             let db = DatabaseSource::new(&database, &host, &password, port, &user);
             let items = db
-                .export(&year, pat_id)
+                .export(
+                    &year,
+                    &ignore_exports_since.unwrap_or("9999-12-31".into()),
+                    pat_id,
+                    include_extern,
+                )
                 .map_err(|_e| "Fehler bei Zugriff auf die Datenbank")?;
 
             let _ = term.clear_last_lines(1);
 
-            let mut writer = Writer::from_path(Path::new(&output)).expect("writeable file");
+            let writer_builder = &mut WriterBuilder::new();
+            let mut writer_builder = writer_builder.has_headers(true);
+            if xls_csv {
+                writer_builder = writer_builder.delimiter(b';');
+            }
+            let mut writer = writer_builder
+                .from_path(Path::new(&output))
+                .expect("writeable file");
 
             items
                 .iter()
@@ -154,6 +201,8 @@ fn main() -> Result<(), Box<dyn Error>> {
                 .green()
                 .to_string(),
             );
+
+            print_extern_notice(include_extern);
         }
         SubCommand::Compare {
             pat_id,
@@ -164,23 +213,11 @@ fn main() -> Result<(), Box<dyn Error>> {
             user,
             file,
             year,
+            ignore_exports_since,
+            include_extern,
         } => {
-            let password = if let Some(password) = password {
-                password
-            } else {
-                let password = dialoguer::Password::new()
-                    .with_prompt("Password")
-                    .interact()
-                    .unwrap_or_default();
-                let _ = term.clear_last_lines(1);
-                password
-            };
-
-            let year = if year.len() == 4 {
-                year
-            } else {
-                format!("2{:0>3}", year)
-            };
+            let password = request_password_if_none(password);
+            let year = sanitize_year(year);
 
             let _ = term.write_line(
                 &style(format!("Warte auf Daten für das Diagnosejahr {}...", year))
@@ -190,7 +227,12 @@ fn main() -> Result<(), Box<dyn Error>> {
 
             let db = DatabaseSource::new(&database, &host, &password, port, &user);
             let db_items = db
-                .export(&year, pat_id)
+                .export(
+                    &year,
+                    &ignore_exports_since.unwrap_or("9999-12-31".into()),
+                    pat_id,
+                    include_extern,
+                )
                 .map_err(|_e| "Fehler bei Zugriff auf die Datenbank")?;
 
             let _ = term.clear_last_lines(1);
@@ -208,6 +250,8 @@ fn main() -> Result<(), Box<dyn Error>> {
                 })
                 .collect::<Vec<_>>();
 
+            print_extern_notice(include_extern);
+
             let _ = term.write_line(
                 &style(format!(
                     "{} Conditions aus der Datenbank für das Jahr {} - aber nicht in Datei '{}'",

src/opal.rs

@@ -41,8 +41,7 @@ impl OpalCsvFile {
 
         let items = reader
             .deserialize::<OpalRecord>()
-            .filter(|record| record.is_ok())
-            .map(|record| record.unwrap())
+            .filter_map(|record| record.ok())
             .map(|record| Record {
                 condition_id: record.cond_id,
                 icd10_code: record.cond_coding_code,
@@ -57,8 +56,7 @@ impl OpalCsvFile {
 
         let items = reader
             .deserialize::<ExportData>()
-            .filter(|record| record.is_ok())
-            .map(|record| record.unwrap())
+            .filter_map(|record| record.ok())
             .collect::<Vec<_>>();
 
         Ok(items)

src/resources/export.sql

@@ -29,22 +29,23 @@ FROM (
         EXTRACTVALUE(lme.xml_daten, '//Patienten_Stammdaten/@Patient_ID') AS pid,
         lme.versionsnummer,
         SHA2(CONCAT('https://fhir.diz.uk-erlangen.de/identifiers/onkostar-xml-condition-id|', EXTRACTVALUE(lme.xml_daten, '//Patienten_Stammdaten/@Patient_ID'), 'condition', EXTRACTVALUE(lme.xml_daten, '//Diagnose/@Tumor_ID')), 256) AS cond_id,
-        SUBSTRING_INDEX(EXTRACTVALUE(lme.xml_daten, '//Primaertumor_ICD_Code'), ' ', 1) AS condcodingcode,
-        SUBSTRING_INDEX(EXTRACTVALUE(lme.xml_daten, '//Diagnosedatum'), ' ', 1) AS diagnosedatum,
-        SUBSTRING_INDEX(SUBSTRING_INDEX(EXTRACTVALUE(lme.xml_daten, '//Diagnosedatum'), ' ', 1), '.', -1) AS diagnosejahr
+        SUBSTRING_INDEX(EXTRACTVALUE(lm.xml_daten, '//Primaertumor_ICD_Code'), ' ', 1) AS condcodingcode,
+        SUBSTRING_INDEX(EXTRACTVALUE(lm.xml_daten, '//Diagnosedatum'), ' ', 1) AS diagnosedatum,
+        SUBSTRING_INDEX(SUBSTRING_INDEX(EXTRACTVALUE(lm.xml_daten, '//Diagnosedatum'), ' ', 1), '.', -1) AS diagnosejahr
     FROM lkr_meldung_export lme
-    JOIN lkr_meldung lm ON (lm.id = lme.lkr_meldung AND lme.typ <> '-1')
-    WHERE lme.xml_daten LIKE '%ICD_Version%'
-        AND SUBSTRING_INDEX(SUBSTRING_INDEX(EXTRACTVALUE(lme.xml_daten, '//Diagnosedatum'), ' ', 1), '.', -1) = :year
-        AND (lme.xml_daten LIKE '%<cTNM%' OR lme.xml_daten LIKE '%<pTNM%' OR lme.xml_daten LIKE '%<Menge_Histologie>%' OR lme.xml_daten LIKE '%<Menge_Weitere_Klassifikation>%')
-
+    JOIN lkr_meldung lm ON (lm.id = lme.lkr_meldung AND lme.typ <> '-1' AND lm.extern <= :include_extern)
+    WHERE lm.xml_daten LIKE '%ICD_Version%'
+        AND SUBSTRING_INDEX(SUBSTRING_INDEX(EXTRACTVALUE(lm.xml_daten, '//Diagnosedatum'), ' ', 1), '.', -1) = :year
+        AND (lm.xml_daten LIKE '%<cTNM%' OR lm.xml_daten LIKE '%<pTNM%' OR lm.xml_daten LIKE '%<Menge_Histologie>%' OR lm.xml_daten LIKE '%<Menge_Weitere_Klassifikation>%')
+        AND (lm.xml_daten NOT LIKE '%histologie_zytologie%')
     ) o1
     LEFT OUTER JOIN (
 
-    SELECT
+    SELECT DISTINCT
         SHA2(CONCAT('https://fhir.diz.uk-erlangen.de/identifiers/onkostar-xml-condition-id|', EXTRACTVALUE(lme.xml_daten, '//Patienten_Stammdaten/@Patient_ID'), 'condition', EXTRACTVALUE(lme.xml_daten, '//Diagnose/@Tumor_ID')), 256) AS cond_id,
-        MAX(versionsnummer) AS max_version
+        CASE WHEN le.exportiert_am < :ignore_exports_since THEN MAX(versionsnummer) ELSE ~0 END AS max_version
     FROM lkr_meldung_export lme
+        JOIN lkr_export le ON (lme.lkr_export = le.id)
     WHERE SUBSTRING_INDEX(SUBSTRING_INDEX(EXTRACTVALUE(lme.xml_daten, '//Diagnosedatum'), ' ', 1), '.', -1) = :year
     GROUP BY cond_id ORDER BY cond_id
 

src/resources/query.sql

@@ -115,21 +115,22 @@ FROM (
         EXTRACTVALUE(lme.xml_daten, '//Patienten_Stammdaten/@Patient_ID') AS pid,
         lme.versionsnummer,
         SHA2(CONCAT('https://fhir.diz.uk-erlangen.de/identifiers/onkostar-xml-condition-id|', EXTRACTVALUE(lme.xml_daten, '//Patienten_Stammdaten/@Patient_ID'), 'condition', EXTRACTVALUE(lme.xml_daten, '//Diagnose/@Tumor_ID')), 256) AS cond_id,
-        SUBSTRING_INDEX(EXTRACTVALUE(lme.xml_daten, '//Primaertumor_ICD_Code'), ' ', 1) AS condcodingcode,
-        SUBSTRING_INDEX(SUBSTRING_INDEX(EXTRACTVALUE(lme.xml_daten, '//Diagnosedatum'), ' ', 1), '.', -1) AS diagnosejahr
+        SUBSTRING_INDEX(EXTRACTVALUE(lm.xml_daten, '//Primaertumor_ICD_Code'), ' ', 1) AS condcodingcode,
+        SUBSTRING_INDEX(SUBSTRING_INDEX(EXTRACTVALUE(lm.xml_daten, '//Diagnosedatum'), ' ', 1), '.', -1) AS diagnosejahr
     FROM lkr_meldung_export lme
-    JOIN lkr_meldung lm ON (lm.id = lme.lkr_meldung AND lme.typ <> '-1')
+    JOIN lkr_meldung lm ON (lm.id = lme.lkr_meldung AND lme.typ <> '-1' AND lm.extern <= :include_extern)
     WHERE lme.xml_daten LIKE '%ICD_Version%'
-        AND SUBSTRING_INDEX(SUBSTRING_INDEX(EXTRACTVALUE(lme.xml_daten, '//Diagnosedatum'), ' ', 1), '.', -1) = :year
-        AND (lme.xml_daten LIKE '%<cTNM%' OR lme.xml_daten LIKE '%<pTNM%' OR lme.xml_daten LIKE '%<Menge_Histologie>%' OR lme.xml_daten LIKE '%<Menge_Weitere_Klassifikation>%')
-
+        AND SUBSTRING_INDEX(SUBSTRING_INDEX(EXTRACTVALUE(lm.xml_daten, '//Diagnosedatum'), ' ', 1), '.', -1) = :year
+        AND (lm.xml_daten LIKE '%<cTNM%' OR lm.xml_daten LIKE '%<pTNM%' OR lm.xml_daten LIKE '%<Menge_Histologie>%' OR lm.xml_daten LIKE '%<Menge_Weitere_Klassifikation>%')
+        AND (lm.xml_daten NOT LIKE '%histologie_zytologie%')
     ) o1
     LEFT OUTER JOIN (
 
-    SELECT
+    SELECT DISTINCT
         SHA2(CONCAT('https://fhir.diz.uk-erlangen.de/identifiers/onkostar-xml-condition-id|', EXTRACTVALUE(lme.xml_daten, '//Patienten_Stammdaten/@Patient_ID'), 'condition', EXTRACTVALUE(lme.xml_daten, '//Diagnose/@Tumor_ID')), 256) AS cond_id,
-        MAX(versionsnummer) AS max_version
+        CASE WHEN le.exportiert_am < :ignore_exports_since THEN MAX(versionsnummer) ELSE ~0 END AS max_version
     FROM lkr_meldung_export lme
+        JOIN lkr_export le ON (lme.lkr_export = le.id)
     WHERE SUBSTRING_INDEX(SUBSTRING_INDEX(EXTRACTVALUE(lme.xml_daten, '//Diagnosedatum'), ' ', 1), '.', -1) = :year
     GROUP BY cond_id ORDER BY cond_id