tlf-qc.qmd

---
title: "QC Report"
format: html
date: "`r Sys.Date()`"
editor: visual
execute:
  keep-md: true
---

```{r, echo = FALSE}
library(waldo)
library(ellmer)
library(stringr)
```

This is the QC report dated `r Sys.Date()`. The report compares the datasets in the `pilot5-submission/pilot5-input/adamdata` directory with the datasets in the `~/GitHub/submissions-pilot3-adam/submission/adam` directory. The datasets are compared using the `diffdf` package.

This run was initiated by `r Sys.getenv("GITHUB_ACTOR")` on the `r Sys.getenv("GITHUB_REF")` Git ref.

```{r session_info}
#| echo: false
sessionInfo()
```

```{r path_setup}
pilot3_out_path <- "pilot5-submission/pilot5-output/pilot3-tlfs/out"
pilot5_out_path <- "pilot5-submission/pilot5-output/pilot5-tlfs/out"
pilot5_out_files <- list.files(pilot5_out_path, pattern = ".out")

pilot3_rtf_path <- "pilot5-submission/pilot5-output/pilot3-tlfs/rtf"
pilot5_rtf_path <- "pilot5-submission/pilot5-output/pilot5-tlfs/rtf"
pilot5_rtf_files <- list.files(pilot5_rtf_path, pattern = ".rtf")

pilot3_pdf_path <- "pilot5-submission/pilot5-output/pilot3-tlfs/pdf"
pilot5_pdf_path <- "pilot5-submission/pilot5-output/pilot5-tlfs/pdf"
pilot5_pdf_files <- list.files(pilot5_pdf_path, pattern = ".pdf")

chat <- chat_anthropic()
# nolint start: line_length_linter
prompt <- "compare these two images and provide a detailed list of any differences found between them? Also, please provide an overall assessment of whether these images appear to be similar enough where a human reviewer would draw the same general conclusion. Finally, give a score from 0-100 of how similar the images are, where 100 is an exact pixel-perfect match and 0 is they are completely different."
# nolint end

type_qc_diff <- type_object(
  difference_list = type_string(),
  same_conclusion = type_boolean(),
  difference_score = type_integer()
)
type_qc_diffs <- type_array(items = type_qc_diff)
```

```{r sink_setup}
sink("qc-tlf.Rmd", append = TRUE)
cat("## TLF QC Check Complete!", "\n\n")
cat("Date: ", format(Sys.Date()), "\n")
cat("Run by: ", Sys.getenv("GITHUB_ACTOR"), "\n")
cat("Git Ref: ", Sys.getenv("GITHUB_REF"), "\n")
cat("BASE: ", "Pilot 5 Datasets", "\n")
cat("COMPARE: ", "Pilot 3 Datasets", "\n")
sink()
```

```{r out_compare}
sink("qc-tlf.Rmd", append = TRUE)

for (i in pilot5_out_files) {
  new_tlf_path <- file.path(pilot5_out_path, i)
  comp_tlf_path <- file.path(pilot3_out_path, sub("pilot5", "pilot3", i))

  diffs <- waldo::compare(readLines(new_tlf_path), readLines(comp_tlf_path))

  if (length(diffs) > 1) file.create("qc-tlf.fail", showWarnings = TRUE)

  cat("<details>\n")
  status_emoji <- if (length(diffs) <= 1) "✅" else "❌"
  cat(str_glue("<summary>{status_emoji} Dataset: {i}</summary>\n\n"))
  cat("\n\n```\n\n")
  print(diffs)
  cat("```\n\n")
  cat("</details>")
  cat("\n\n")
}
sink()
```

```{r pdf_compare}
sink("qc-tlf.Rmd", append = TRUE)

for (i in pilot5_pdf_files) {
  new_tlf_path <- file.path(pilot5_pdf_path, i)
  new_png_path <- sub(".pdf", ".png", new_tlf_path, fixed = TRUE)

  comp_tlf_path <- file.path(pilot3_pdf_path, sub("pilot5", "pilot3", i))
  comp_png_path <- sub(".pdf", ".png", comp_tlf_path, fixed = TRUE)

  system(
    paste0("convert -adjoin ", new_tlf_path, " ", new_png_path),
    intern = TRUE,
    ignore.stdout = TRUE,
    ignore.stderr = TRUE,
    wait = TRUE
  )
  system(
    paste0("convert -adjoin ", comp_tlf_path, " ", comp_png_path),
    intern = TRUE,
    ignore.stdout = TRUE,
    ignore.stderr = TRUE,
    wait = TRUE
  )

  new_image <- content_image_file(new_png_path)
  comp_image <- content_image_file(comp_png_path)

  res <- chat$chat_structured(
    prompt,
    new_image,
    comp_image,
    type = type_qc_diffs
  )

  if (!res$same_conclusion) file.create("qc-tlf.fail", showWarnings = TRUE)

  cat("<details>\n")
  status_emoji <- if (res$same_conclusion) "✅" else "❌"
  cat(str_glue("<summary>{status_emoji} TLF: {i} - Score: {res$difference_score}</summary>\n\n"))
  cat("\n\n")
  print(res$difference_list[[1]])
  cat("\n\n")
  cat("</details>")
  cat("\n\n")
}

sink()
```

```{r rtf_compare}
sink("qc-tlf.Rmd", append = TRUE)

for (i in pilot5_rtf_files) {
  new_tlf_path <- file.path(pilot5_rtf_path, i)
  new_png_path <- sub(".rtf", ".png", new_tlf_path, fixed = TRUE)
  new_pdf_path <- sub(".rtf", ".pdf", new_tlf_path, fixed = TRUE)

  comp_tlf_path <- file.path(pilot3_rtf_path, sub("pilot5", "pilot3", i))
  comp_png_path <- sub(".rtf", ".png", comp_tlf_path, fixed = TRUE)
  comp_pdf_path <- sub(".rtf", ".pdf", comp_tlf_path, fixed = TRUE)


  Sys.setenv("LD_LIBRARY_PATH" = paste0("/usr/lib/libreoffice/program:", Sys.getenv("LD_LIBRARY_PATH")))
  system(
    paste0(
      "libreoffice --headless --convert-to pdf --outdir ",
      pilot5_rtf_path, " ", new_tlf_path
    ),
    intern = TRUE,
    ignore.stdout = FALSE,
    ignore.stderr = FALSE,
    wait = TRUE
  )

  system(
    paste0(
      "libreoffice --headless --convert-to pdf --outdir ",
      pilot3_rtf_path, " ", comp_tlf_path
    ),
    intern = TRUE,
    ignore.stdout = FALSE,
    ignore.stderr = FALSE,
    wait = TRUE
  )

  system(
    paste0("convert -adjoin -append ", new_pdf_path, " ", new_png_path),
    intern = TRUE,
    ignore.stdout = FALSE,
    ignore.stderr = FALSE,
    wait = TRUE
  )
  system(
    paste0("convert -adjoin -append ", comp_pdf_path, " ", comp_png_path),
    intern = TRUE,
    ignore.stdout = TRUE,
    ignore.stderr = TRUE,
    wait = TRUE
  )

  new_image <- content_image_file(new_png_path)
  comp_image <- content_image_file(comp_png_path)

  res <- chat$chat_structured(
    prompt,
    new_image,
    comp_image,
    type = type_qc_diffs
  )

  if (!res$same_conclusion) file.create("qc-tlf.fail", showWarnings = TRUE)
  cat("<details>\n")
  status_emoji <- if (res$same_conclusion) "✅" else "❌"
  cat(str_glue("<summary>{status_emoji} TLF: {i} - Score: {res$difference_score}</summary>\n\n"))
  cat("\n\n")
  print(res$difference_list[[1]])
  cat("\n\n")
  cat("</details>")
  cat("\n\n")
}

sink()
```