Additional config param for support guess envs, update fallback type & more docs (#158)

Author: robherley

docs/self-hosting.md

@@ -10,6 +10,7 @@ The architecture of snips.sh was designed with self-hosting in mind, it's very s
     - [Host Keys](#host-keys)
     - [Limiting SSH Access](#limiting-ssh-access)
     - [Statsd Metrics](#statsd-metrics)
+    - [Build without Tensorflow](#build-without-tensorflow)
   - [Examples](#examples)
     - [Docker Compose](#docker-compose)
 
@@ -143,6 +144,17 @@ ssh-import-id gh:robherley -o snips_authorized_keys
 
 At runtime, snips.sh will emit various metrics if the `SNIPS_METRICS_STATSD` is defined. This should be the full udp address with the protocol, e.g. `udp://localhost:8125`.
 
+### Build without Tensorflow
+
+In order to "guess" what language a snippet is, snips.sh will use [guesslang](https://github.com/yoeo/guesslang) (specifically [guesslang-go](https://github.com/robherley/guesslang-go)). This requires the [`libtensorflow`](https://www.tensorflow.org/install/lang_c) C API.
+
+If you do not want tensorflow as a runtime dependency (or if your environment does not support it) you can build without the guesser activated with the `noguesser` build tag:
+
+```
+go build -tags noguesser
+```
+
+For `arm64` systems, this is automatic and will default to not compiling with the guesslang dependency.
 
 ## Examples
 

internal/config/config.go

@@ -5,7 +5,6 @@ import (
 	"fmt"
 	"net/url"
 	"os"
-	"runtime"
 	"text/tabwriter"
 	"time"
 
@@ -127,7 +126,7 @@ func Load() (*Config, error) {
 		return nil, err
 	}
 
-	cfg.EnableGuesser = cfg.EnableGuesser && runtime.GOARCH == "amd64"
+	cfg.EnableGuesser = cfg.EnableGuesser && GuessingSupported
 
 	return cfg, nil
 }

internal/config/config_guesser.go

@@ -0,0 +1,5 @@
+//go:build amd64 && !noguesser
+
+package config
+
+const GuessingSupported = true

internal/config/config_noguesser.go

@@ -0,0 +1,6 @@
+//go:build arm64 || noguesser
+
+package config
+
+// https://github.com/robherley/snips.sh/issues/39
+const GuessingSupported = false

internal/renderer/guess.go

@@ -1,9 +1,34 @@
-//go:build arm64 || noguesser
+//go:build amd64 && !noguesser
 
 package renderer
 
+import (
+	"strings"
+	"time"
+
+	"github.com/armon/go-metrics"
+	"github.com/robherley/guesslang-go/pkg/guesser"
+	"github.com/rs/zerolog/log"
+)
+
+var guesslang *guesser.Guesser
+
+func init() {
+	var err error
+	guesslang, err = guesser.New()
+	if err != nil {
+		log.Fatal().Err(err).Msg("failed to initialize guesslang")
+	}
+}
+
 func Guess(content string) string {
-	// currently not supporting guessing in arm64 bc of libtensorflow requirements
-	// https://github.com/robherley/snips.sh/issues/39
-	panic("not implemented")
+	guessStart := time.Now()
+	answer, err := guesslang.Guess(content)
+	metrics.MeasureSince([]string{"guess", "duration"}, guessStart)
+	if err != nil {
+		log.Warn().Err(err).Msg("failed to guess the file type")
+		return ""
+	}
+
+	return strings.ToLower(answer.Predictions[0].Language)
 }

internal/renderer/guess_amd64.go

@@ -0,0 +1,9 @@
+//go:build arm64 || noguesser
+
+package renderer
+
+func Guess(content string) string {
+	// currently not supporting guessing in arm64 bc of libtensorflow requirements
+	// https://github.com/robherley/snips.sh/issues/39
+	panic("not implemented")
+}

internal/renderer/guess_disabled.go

@@ -5,9 +5,11 @@ import (
 	"github.com/alecthomas/chroma/v2/lexers"
 )
 
-var (
-	FallbackLexer = lexers.Fallback
-)
+var FallbackLexer chroma.Lexer = chroma.MustNewLexer(&chroma.Config{
+	Name:      "plaintext",
+	Filenames: []string{"*"},
+	Priority:  -1,
+}, lexers.PlaintextRules)
 
 func Analyze(content string) chroma.Lexer {
 	lexer := lexers.Analyse(content)