prefixes: Add 'resolver-blocked.txt' config

cmd/bootstrap.go

@@ -143,7 +143,7 @@ func (bs bootstrap) Run(context wisski_distillery.Context) error {
 					core.DefaultOverridesJSON,
 					fs.ModePerm,
 				); err != nil {
-					return errBootstrapCreateFile.WithMessageF(err)
+					return err
 				}
 
 				context.Println(tpl.AuthorizedKeys)
@@ -153,12 +153,22 @@ func (bs bootstrap) Run(context wisski_distillery.Context) error {
 					core.DefaultAuthorizedKeys,
 					fs.ModePerm,
 				); err != nil {
-					return errBootstrapCreateFile.WithMessageF(err)
+					return err
+				}
+
+				context.Println(tpl.SelfResolverBlockFile)
+				if err := environment.WriteFile(
+					env,
+					tpl.SelfResolverBlockFile,
+					core.DefaultResolverBlockedTXT,
+					fs.ModePerm,
+				); err != nil {
+					return err
 				}
 
 				return nil
 			}, context.IOStream, "Creating additional config files"); err != nil {
-				return err
+				return errBootstrapCreateFile.WithMessageF(err)
 			}
 		}
 

cmd/update_prefix_config.go

@@ -53,7 +53,9 @@ func (upc updateprefixconfig) Run(context wisski_distillery.Context) error {
 			// read the prefix config
 			data, err := instance.PrefixConfig()
 			if err != nil {
-				return err
+				data = "# error, skipped\n"
+				context.EPrintln(err)
+				err = nil
 			}
 			context.IOStream.Printf("%s", data)
 

internal/component/control/backup.go

@@ -31,6 +31,7 @@ func (control *Control) backupFiles() []string {
 		control.Config.ConfigPath,
 		control.Config.ExecutablePath(),
 		control.Config.SelfOverridesFile,
+		control.Config.SelfResolverBlockFile,
 		control.Config.GlobalAuthorizedKeysFile,
 	}
 }

internal/component/control/control.env

@@ -7,3 +7,4 @@ CONFIG_PATH=${CONFIG_PATH}
 DEPLOY_ROOT=${DEPLOY_ROOT}
 GLOBAL_AUTHORIZED_KEYS_FILE=${GLOBAL_AUTHORIZED_KEYS_FILE}
 SELF_OVERRIDES_FILE=${SELF_OVERRIDES_FILE}
+SELF_RESOLVER_BLOCK_FILE=${SELF_RESOLVER_BLOCK_FILE}

internal/component/control/control.go

@@ -46,6 +46,7 @@ func (control *Control) Stack(env environment.Environment) component.StackWithRe
 
 			"GLOBAL_AUTHORIZED_KEYS_FILE": control.Config.GlobalAuthorizedKeysFile,
 			"SELF_OVERRIDES_FILE":         control.Config.SelfOverridesFile,
+			"SELF_RESOLVER_BLOCK_FILE":    control.Config.SelfResolverBlockFile,
 		},
 
 		TouchFiles:       []string{control.ResolverFile},

internal/component/control/control/docker-compose.yml

@@ -22,6 +22,7 @@ services:
       - "${DEPLOY_ROOT}:${DEPLOY_ROOT}:ro"
       - "${GLOBAL_AUTHORIZED_KEYS_FILE}:${GLOBAL_AUTHORIZED_KEYS_FILE}:ro"
       - "${SELF_OVERRIDES_FILE}:${SELF_OVERRIDES_FILE}:ro"
+      - "${SELF_RESOLVER_BLOCK_FILE}:${SELF_RESOLVER_BLOCK_FILE}:ro"
 
 networks:
   default:

internal/component/instances/wisski_prefix.go

@@ -1,7 +1,7 @@
 package instances
 
 import (
-	"errors"
+	"bufio"
 	"io"
 	"path/filepath"
 	"strings"
@@ -25,7 +25,7 @@ var listURIPrefixesPHP string
 // Prefixes returns the prefixes
 func (wisski *WissKI) Prefixes() (prefixes []string, err error) {
 	// get all the ugly prefixes
-	err = wisski.ExecPHPScript(stream.FromEnv(), &prefixes, listURIPrefixesPHP, "list_prefixes")
+	err = wisski.ExecPHPScript(stream.FromNil(), &prefixes, listURIPrefixesPHP, "list_prefixes")
 	if err != nil {
 		return nil, err
 	}
@@ -35,29 +35,53 @@ func (wisski *WissKI) Prefixes() (prefixes []string, err error) {
 		return strings.HasPrefix(now, prev)
 	})
 
+	// load the list of blocked prefixes
+	blocks, err := wisski.instances.BlockedPrefixes()
+	if err != nil {
+		return nil, err
+	}
+
 	// filter out blocked prefixes
-	return slicesx.Filter(prefixes, func(uri string) bool { return !IsNonServedURI(uri) }), nil
+	return slicesx.Filter(prefixes, func(uri string) bool { return !hasAnyPrefix(uri, blocks) }), nil
 }
 
-// TODO: Eventually move this into a configuration file.
-// But for now this is fine
-var blockedURIs = []string{
-	"http://erlangen-crm.org/",
-	"http://www.w3.org/",
-	"xsd:",
+func (instances *Instances) BlockedPrefixes() ([]string, error) {
+	// open the resolver block file
+	file, err := instances.Environment.Open(instances.Config.SelfResolverBlockFile)
+	if err != nil {
+		return nil, err
 	}
 
-func IsNonServedURI(candidate string) bool {
+	var lines []string
+
+	// read all the lines that aren't a comment!
+	scanner := bufio.NewScanner(file)
+	for scanner.Scan() {
+		line := strings.TrimSpace(scanner.Text())
+		if line == "" || strings.HasPrefix(line, "//") || strings.HasPrefix(line, "#") {
+			continue
+		}
+		lines = append(lines, line)
+	}
+
+	// check if there was an error
+	if err := scanner.Err(); err != nil {
+		return nil, err
+	}
+
+	// and done!
+	return lines, nil
+}
+
+func hasAnyPrefix(candidate string, prefixes []string) bool {
 	return slicesx.Any(
-		blockedURIs,
+		prefixes,
 		func(prefix string) bool {
 			return strings.HasPrefix(candidate, prefix)
 		},
 	)
 }
 
-var errPrefixExecFailed = errors.New("PrefixConfig: Failed to call list_uri_prefixes")
-
 // PrefixConfig returns the prefix config belonging to this instance.
 func (wisski *WissKI) PrefixConfig() (config string, err error) {
 	// if the user requested to skip the prefix, then don't do anything with it!

internal/config/config.go

@@ -36,6 +36,10 @@ type Config struct {
 	// Do this by adding URLs (without trailing '/'s) into a JSON file
 	SelfOverridesFile string `env:"SELF_OVERRIDES_FILE" default:"" parser:"file"`
 
+	// You can block specific prefixes from being picked up by the resolver.
+	// Do this by adding one prefix per file.
+	SelfResolverBlockFile string `env:"SELF_RESOLVER_BLOCK_FILE" default:"" parser:"file"`
+
 	// The system can support setting up certificate(s) automatically.
 	// It can be enabled by setting an email for certbot certificates.
 	// This email address can be configured here.

internal/config/config_template

@@ -19,6 +19,10 @@ SELF_EXTRA_DOMAINS=
 # Do this by adding URLs (without trailing '/'s) into a JSON file
 SELF_OVERRIDES_FILE=${SELF_OVERRIDES_FILE}
 
+# You can block specific prefixes within Triplestore from showing up in the resolver configuration file.
+# Do this by adding one prefix per line in this file.
+SELF_RESOLVER_BLOCK_FILE=${SELF_RESOLVER_BLOCK_FILE}
+
 # The system can support setting up certificate(s) automatically. 
 # It can be enabled by setting an email for certbot certificates. 
 # This email address can be configured here. 
@@ -28,7 +32,6 @@ CERTBOT_EMAIL=
 # Backups older than this will be removed when a new backup is made.
 MAX_BACKUP_AGE=30
 
-
 # Each Drupal instance requires a corresponding system user, database users and databases. 
 # These are also set by the appropriate domain name. 
 # To differentiate them from other users of the system, these names can be prefixed. 

internal/config/template.go

@@ -20,6 +20,7 @@ type Template struct {
 	DeployRoot               string `env:"DEPLOY_ROOT"`
 	DefaultDomain            string `env:"DEFAULT_DOMAIN"`
 	SelfOverridesFile        string `env:"SELF_OVERRIDES_FILE"`
+	SelfResolverBlockFile    string `env:"SELF_RESOLVER_BLOCK_FILE"`
 	AuthorizedKeys           string `env:"AUTHORIZED_KEYS_FILE"`
 	TriplestoreAdminUser     string `env:"GRAPHDB_ADMIN_USER"`
 	TriplestoreAdminPassword string `env:"GRAPHDB_ADMIN_PASSWORD"`
@@ -43,6 +44,10 @@ func (tpl *Template) SetDefaults(env environment.Environment) (err error) {
 		tpl.SelfOverridesFile = filepath.Join(tpl.DeployRoot, core.OverridesJSON)
 	}
 
+	if tpl.SelfResolverBlockFile == "" {
+		tpl.SelfResolverBlockFile = filepath.Join(tpl.DeployRoot, core.ResolverBlockedTXT)
+	}
+
 	if tpl.AuthorizedKeys == "" {
 		tpl.AuthorizedKeys = filepath.Join(tpl.DeployRoot, core.AuthorizedKeys)
 	}

internal/core/bootstrap/resolver-blocked.txt

@@ -0,0 +1,16 @@
+# This file contains prefixes that should not be picked up by the global resolver.
+# They will not appear in the list of prefixes.
+# It should be one prefix per line, '#' and '//' as well as blank lines are treated as comments
+
+# definitely shouldn't be resolved to any WissKI
+http://www.w3.org/
+xsd:
+urn:
+
+# generic prefixes by some common adapters.
+# you may or may not want these.
+http://d-nb.info/gnd/
+http://erlangen-crm.org/
+http://id.gnm.de/ont/
+http://zotero.org/
+https://api.zotero.org/

internal/core/core.go

@@ -23,6 +23,14 @@ const OverridesJSON = "overrides.json"
 //go:embed bootstrap/overrides.json
 var DefaultOverridesJSON []byte
 
+// ResolverBlockTXT is the name of the resolver blocked prefix file.
+// It should be located inside the deployment directory.
+const ResolverBlockedTXT = "resolver-blocked.txt"
+
+// ResolverBlockTXT contains a template for 'resolver-blocked' file
+//go:embed bootstrap/resolver-blocked.txt
+var DefaultResolverBlockedTXT []byte
+
 // AuthorizedKeys contains the default name for the 'global_authorized_keys' file
 const AuthorizedKeys = "authorized_keys"